CN113239439B - Shield construction ground surface settlement prediction system and method - Google Patents

Abstract

The invention discloses a system and a method for predicting shield construction surface subsidence, wherein the system for predicting shield construction surface subsidence comprises: the meta-attribute extraction module is used for recombining the original data sets to extract meta-attributes and calculating each attribute characteristic index; the settlement data generator training module is used for constructing a settlement data generator based on the meta-attribute; the settlement data generation module is used for generating a group of simulation data by combining the engineering characteristics of the current construction project; the settlement prediction model pre-training module is used for training a settlement prediction model of the current engineering by combining the generated simulation data to obtain an initial settlement prediction model; and the real-time settlement prediction module is used for acquiring real-time shield tunneling data and predicting the settlement value of the ground surface settlement monitoring point. The system and the method for predicting the shield construction surface subsidence can improve the applicability and accuracy of prediction without large data accumulation in the early period.

Description

Prediction system and method of ground settlement for shield construction

technical field

The invention belongs to the technical field of surface settlement prediction, relates to a surface settlement prediction system, and in particular relates to a surface settlement prediction system and method for shield construction.

Background technique

The shield method can be applied to construction under various hydrogeological conditions, and the number of shields in the world is increasing year by year because of its stronger adaptability to the stratum. The shield method has been widely used in underground engineering. During the construction of the shield method, the soil layer around the tunnel will be deformed, especially in the soft aquifer or unstable soil layer. Since the construction site of the shield method is mainly located in the prosperous part of the city, the settlement control of the surrounding soil is very important. Once the settlement exceeds the control range, it may affect the safety of surrounding buildings. It can be seen that it is very meaningful to study the real-time prediction technology of land subsidence in the process of shield tunnel excavation.

In the actual construction process, the geometric characteristics of the tunnel, the characteristics of the excavated soil layer, the setting of the shield parameters, and the disturbance of the soil will greatly affect the stratum displacement caused by the advancement of the shield. This is a complex combined effect of various factors. the result of. Moreover, it is usually difficult to obtain accurate geological parameters due to the complex geological conditions on site, and the relationship between the low-frequency subsidence monitoring data and the high-frequency shield tunneling data is difficult to clarify. The high-dimensional, high-frequency and imprecise characteristics of shield construction data pose challenges to the establishment of accurate and universally applicable settlement prediction models.

To achieve real-time prediction of land subsidence in the process of shield tunnel construction, it is necessary to establish an effective prediction model, and to clarify the mathematical relationship between shield excavation parameters, excavation environment and land subsidence. However, the existing construction data generally have certain errors, and the high-dimensional and high-frequency shield tunneling parameter data and the low-frequency and large-spacing land subsidence data also cause great difficulty in establishing a settlement prediction model. The existing settlement prediction methods can be summarized into the following three types: 1) empirical formula method. Patent CN201610707226.4 proposes a method for predicting the surface settlement caused by the entry and discharge of earth pressure balance shield soil bins. This method obtains the relationship between shield tunneling parameters and surface deformation through formula derivation, but there are problems with this method. It is because the parameters cannot be determined, the shape of the settlement tank is fixed, and it is impossible to accurately predict the surface settlement in various environments. 2) Settlement feedback method. Patent CN201910068462.X proposes a method for predicting ground settlement of shield construction based on cyclic neural network. This method predicts the settlement of the next stage based on the previous settlement value, which can only be used in the stage when the project is in a stable state and can only be used for short-term Prediction, inability to adapt to changes in environment and propulsion parameters. 3) Tunneling parameter method. Patent CN201810650131.2 proposes a method for predicting land subsidence during shield construction based on dual-model fusion, and patent CN201710344450.6 proposes a method for predicting land subsidence caused by shield construction based on a neuro-fuzzy inference system. Predicting settlement requires high requirements on geological environment, burial depth, and equipment similarity, and has poor generalization ability, making it difficult to popularize.

In view of this, there is an urgent need to design a new method for predicting the surface settlement of tunnel construction in order to overcome at least some of the above-mentioned shortcomings of the existing methods for predicting the surface settlement of tunnel construction.

SUMMARY OF THE INVENTION

The invention provides a system and method for predicting ground settlement of shield construction, which can improve the applicability and accuracy of prediction, and does not require a large amount of data accumulation in the early stage.

In order to solve the above-mentioned technical problems, according to one aspect of the present invention, the following technical solutions are adopted:

A shield construction surface settlement prediction system, the shield construction surface settlement prediction system includes:

The meta-attribute extraction module is used to recombine the original data set to extract the meta-attributes, and calculate the characteristic indicators of each attribute;

The subsidence data generator training module is used to construct a meta-attribute-based subsidence data generator;

The settlement data generation module is used to generate a set of simulation data in combination with the engineering characteristics of the current construction project;

The settlement prediction model pre-training module is used to train the settlement prediction model of the current project in combination with the generated simulation data to obtain the initial settlement prediction model;

The real-time settlement prediction module is used to obtain real-time shield tunneling data to predict the settlement value of the surface settlement monitoring point.

As an embodiment of the present invention, the element attribute extraction module is used to extract element attributes; the main factors affecting the surface subsidence are divided into three attribute categories: foundation category, disturbance category and subsidence category;

Each type of meta-attribute includes a variety of excavation characteristics, which are calculated from the original construction data. The specific calculation methods are as follows:

The basic category is the basic information required for the construction of the shield tunnel, including soil properties, geometric properties and technological properties;

The soil properties specifically include the cohesion, internal friction angle and moisture content of each ring tunnel section. The specific calculation methods of each ring soil property are as follows:

Step S111: Calculate the ratio P _m of each layer of soil in the ring to the excavation surface area respectively, define the buried depth of the upper boundary of the m-th soil layer as d _1,m , the buried depth of the lower boundary as d _2,m , and the buried depth of the center point of the tunnel section. The depth is d, the radius of the excavation surface is R, S _{1, m} is the contact area between the soil body and the excavation surface above the starting elevation of the soil layer, and S _{2, m} is the contact area between the soil body and the excavation surface above the end elevation of the soil layer Area; specific calculation methods include:

Step S112: Calculate the soil properties of the ring tunnel section, define the cohesion, internal friction angle and water content of the m-th soil layer as C _m , φ _m and ω _m , respectively. The friction angle and water content are C, φ and ω, respectively, and the calculation formulas are as follows:

C=∑P _m ·C _m ϕ= _∑Pm · _ϕm ω= _∑Pm · _ωm

The geometric characteristics include the buried depth of each ring of the shield and the soil loss rate in the case of non-grouting. The buried depth of each ring of the shield can be obtained directly from the tunnel design document; the soil loss rate of each ring in the case of non-grouting needs to be obtained. The straight line segment and the curved segment are calculated independently, and the straight line segment only considers the volume of the ring enclosed by the radius R of the shield cutter head and the outer radius r of the tunnel segment. The specific calculation formula is as follows:

υ=(R ² -r ² )πl

In order to deflect the shield during the advancement of the curved section, the over-excavation of the soil is usually caused. The theoretical soil loss rate is quite different from that of the straight section. The specific calculation method is as follows:

In the formula, R ₀ tunnel curvature radius, l is the segment width, L is the length of the shield machine, D is the diameter of the shield machine, δ is the over-excavation inside the shield machine, δ′ is the theoretical gap in the middle of the shield machine, and δ′≈δ;

The process characteristics include the type of shield machine, the length of the shield and the diameter of the shield. The types of shield machines are mainly divided into earth pressure balance shield machines and mud-water balance shield machines, which are determined according to the type of shield machine selected for the actual project. ; Shield length and shield diameter are obtained according to the parameter data of the shield machine used in the specific construction;

Disturbance parameters are mainly indicators for evaluating the degree of soil disturbance caused by the shield during the excavation process, which are obtained by analyzing and processing the real-time excavation data of the shield; they are divided into notch disturbance characteristics, attitude disturbance characteristics and shield tail according to the different disturbance mechanisms. Disturbance characteristics:

The theoretical pressure difference of shield tunnel is used to evaluate the disturbance of the soil in front of the incision, so as to obtain the disturbance characteristics of the incision. The specific calculation method is as follows:

Step S121, calculating the upper load size Q according to the geological exploration report;

Step S122, calculating the lateral earth pressure coefficient K of the soil body on the excavation surface according to the tunnel geological survey report;

Step S123, extracting the earth pressure data during the advancing period of the ring, and calculating the average earth pressure P of the ring;

Step S124, calculate the theoretical earth pressure difference according to the following formula: ΔP=P-Q·K;

The attitude disturbance characteristics include the attitude change of the shield in the horizontal direction and the attitude change in the elevation direction of the shield; the attitude change in the elevation direction of the shield is represented by the change in the pitch angle of the shield, namely:

为盾构环俯仰角变化量，

为每环开始时的俯仰角，

为每环结束时的俯仰角；in the formula

is the variation of the pitch angle of the shield ring,

is the pitch angle at the beginning of each loop,

is the pitch angle at the end of each loop;

The attitude change of the shield in the horizontal direction is represented by the change of the horizontal angle, which is calculated according to the deviation of the shield and the linearity of the axis. The specific calculation method is as follows:

In the formula, h' _x is the horizontal deviation of the incision at the beginning of each ring, h" _x is the horizontal deviation of the incision at the end of each ring, t' _x is the horizontal deviation of the shield tail at the beginning of each ring, and t" _x is the horizontal deviation of each ring. The amount of horizontal deviation of the shield tail at the end of the ring;

The grouting pressure of the shield tail is obtained by calculating the average pressure of the shield tail grouting in each ring, and the filling rate of the shield tail grouting is calculated according to the ratio of the cumulative grouting amount of the ring to the theoretical shield tail gap, so as to obtain the shield tail disturbance characteristics. The formula is as follows:

In the formula, g is the filling rate of the shield tail grouting, G _t is the cumulative grouting amount of the shield tail ring, and v is the theoretical shield tail gap size;

The subsidence category mainly describes the subsidence in the process of shield tunneling through the subsidence form and subsidence amplitude;

The settlement form is mainly represented by the width coefficient of the lateral settlement tank. The width coefficient I of the lateral settlement tank is obtained by using the peck formula to fit the curve of the lateral settlement measuring points within the influence range and take the average value. The peck formula is as follows:

where S is the settlement value of any point on the surface; S _max is the maximum surface settlement, located just above the tunnel axis; x is the horizontal distance between any point and the tunnel axis; I is the settlement tank width coefficient, that is, the tunnel axis and the settlement tank the distance of the inflection point;

The subsidence range includes the average subsidence 10m in front of the incision, the average subsidence above the shield body, and the average subsidence 10m behind the shield tail.

As an embodiment of the present invention, the training module of the settlement data generator is used to extract the meta-attributes from the acquired construction data of different projects according to the above method, and then use the foundation class and disturbance class attributes as model inputs, and the settlement class attributes As the model output, a recurrent neural network model is trained as a settlement data generator; the specific steps are as follows:

Step S21, data preprocessing, convert the meta-attribute data into time series data, and divide the training set and the test set;

Step S22, constructing a three-layer cyclic neural network model;

Step S23, using the training set data for model training, and verifying the model prediction effect on the test set, and adjusting the model parameters according to the results;

Step S24, save the model;

The subsidence data generation module is used to calculate the basic data according to the geometric data, geological data and shield machine parameter data of the actual application project, and then calculate the disturbance data range in combination with the theoretical calculation method;

When calculating the notch disturbance characteristics, the earth pressure fluctuation in the actual propulsion process is simulated according to the theoretical earth pressure value plus a certain random fluctuation amount, and then the simulated theoretical earth pressure difference is calculated; when calculating the shield tail disturbance characteristics, the average grouting The pressure is simulated according to the grouting pressure range under similar soil layers, and the average grouting rate fluctuates up and down according to the theoretical grouting filling rate of 150%; when calculating the attitude disturbance characteristics, the theoretical shield tunnel is linearly obtained according to the design axis of the shield tunnel Elevation and horizontal attitude changes.

Input the simulated construction data generated by the above method into the trained settlement data generator to obtain the settlement attributes at each moment; obtain the settlement amount of each point with the help of the empirical settlement curve formula according to the calculated settlement attributes. The specific steps are: as follows:

Step S31, obtaining the actual construction project tunnel geometric data, geological data and shield machine parameter data;

Step S32, designing a simulated construction scheme according to the basic data;

Step S33, input the simulated construction data into the settlement data generator to obtain settlement data;

Step S34: Calculate the settlement of each settlement monitoring point at each moment according to the settlement data, and the specific calculation method is as follows:

Firstly, according to the calculated settlement data combined with the empirical longitudinal settlement curve formula, the longitudinal settlement curve is obtained. The empirical formula is as follows:

In the formula, S _y is the settlement of the settlement monitoring point at the distance x from the incision on the axis, y is the distance between the monitoring point and the incision, and y is a negative number when the monitoring point is in front of the incision;

Then combined with the lateral settlement tank coefficient I, the lateral settlement tank formula of the settlement monitoring point can be calculated:

Finally, by combining the above two formulas, the settlement value of the settlement monitoring point whose relative incision coordinates are (x, y) can be calculated.

Step S35, save the settlement simulation data.

As an embodiment of the present invention, the settlement prediction model pre-training module is used to construct the LSTM monitoring point settlement prediction model according to the shield construction simulation data calculated by the settlement data generator. Soil cohesion on the excavation face, friction angle in the soil on the excavation face, moisture content of the soil on the excavation face, upper load, frontal earth pressure, grouting filling rate, grouting pressure, change in the horizontal attitude of the shield, shield The attitude change in the elevation direction, the longitudinal distance between the settlement monitoring point and the incision, and the lateral distance between the settlement monitoring point and the axis; the model output is the cumulative settlement value at time t; among them, the single sample time step n of the LSTM model is 12, and the LSTM layer the number is 1;

The real-time settlement prediction module is used to predict the settlement of the settlement monitoring points within the influence range of the shield by adopting the settlement prediction model, and specifically includes the following steps:

Step S51, obtaining the current shield position, and determining a set of monitoring points for which settlement prediction needs to be performed, specifically including horizontal and vertical monitoring points within the range of ten meters in front of the incision to ten meters behind the shield tail;

Step S52, obtaining historical shield construction data from time t-n to time t;

Step S53, obtaining the setting values of the shield construction parameters of the next 5 time steps;

Step S54, input the shield construction data from time t-n+i to time t+i, and use the LSTM model to predict the settlement of each monitoring point at time t+i;

Step S55, repeating step S54 until the settlement prediction for the next five time steps is completed.

As an embodiment of the present invention, the system further includes:

The data acquisition module of historical construction projects is used to collect shield construction data, excavation parameter data, geological parameter data and land subsidence data under different construction environments;

The settlement prediction model automatic learning module is used to update the settlement prediction model online.

The historical construction project data acquisition module is used to acquire shield construction data of different engineering projects from the shield construction database, classify the acquired original construction data, extract the data related to settlement prediction, and generate an original data set ; The original data set specifically includes geological data set, tunnel geometry data set, shield machine parameter data set, shield excavation data set and subsidence data set. The data contained in each type of data set are as follows:

The geological data set includes: drilling point position, starting and ending depth of each soil layer, soil mechanical parameters of each soil layer; soil mechanical parameters of each soil layer include cohesion, internal friction angle, compressibility, compressive modulus, natural pores ratio, lateral earth pressure coefficient, natural moisture content, severity;

The tunnel geometry dataset includes: the outer diameter of the tunnel segment, the buried depth of the tunnel, and the line type of the tunnel;

Shield machine parameter data set includes: shield diameter, shield length, shield process type;

The shield tunneling data set includes: shield tunneling process, cut horizontal deviation, cut elevation deviation, shield tail horizontal deviation, shield tail elevation deviation, shield slope angle, frontal earth pressure (zone), grouting amount (distribution), Slurry pressure (distribution), tunneling speed, total thrust, slurry type (single-fluid/double-fluid), slurry initial setting time;

The settlement data set includes: measuring point location, monitoring time, settlement value;

The settlement prediction model automatic learning module is used for rolling training the settlement prediction model according to the new data generated on the construction site, so as to improve the accuracy of the module calculation;

The subsidence prediction model is updated in the background by inputting the shield parameter data newly generated from the last training to the module every K _t time, so as to improve the calculation accuracy of the model.

According to another aspect of the present invention, the following technical scheme is adopted: a method for predicting surface settlement of shield tunnel construction, the method for predicting surface settlement of shield tunnel construction includes:

The meta-attribute extraction step is to recombine the original data set to extract the meta-attributes, and calculate the characteristic indicators of each attribute;

The subsidence data generator training step is to construct a meta-attribute-based subsidence data generator;

The step of generating settlement data, combining with the engineering characteristics of the current construction project, generates a set of simulated data;

In the pre-training step of the settlement prediction model, the settlement prediction model of the current project is trained in combination with the generated simulation data, and the initial settlement prediction model is obtained;

The real-time settlement prediction step is to obtain real-time shield tunneling data to predict the settlement value of the surface settlement monitoring point.

As an embodiment of the present invention, the method further includes:

Data acquisition steps of historical construction projects, collecting shield construction data, excavation parameter data, geological parameter data and land subsidence data under different construction environments;

The automatic learning step of the settlement prediction model, and the online update of the settlement prediction model;

In the historical construction project data acquisition step, the shield construction data of different engineering projects is acquired from the shield construction database, the acquired original construction data is classified, the data related to settlement prediction is extracted, and an original data set is generated. ; The original data set specifically includes geological data set, tunnel geometry data set, shield machine parameter data set, shield excavation data set and subsidence data set. The data contained in each type of data set are as follows:

The geological data set includes: drilling point position, starting and ending depth of each soil layer, soil mechanical parameters of each soil layer; soil mechanical parameters of each soil layer include cohesion, internal friction angle, compressibility, compressive modulus, natural pores ratio, lateral earth pressure coefficient, natural moisture content, severity;

The tunnel geometry dataset includes: the outer diameter of the tunnel segment, the buried depth of the tunnel, and the line type of the tunnel;

Shield machine parameter data set includes: shield diameter, shield length, shield process type;

The shield tunneling data set includes: shield tunneling process, cut horizontal deviation, cut elevation deviation, shield tail horizontal deviation, shield tail elevation deviation, shield slope angle, frontal earth pressure (zone), grouting amount (distribution), Slurry pressure (distribution), tunneling speed, total thrust, slurry type (single-fluid/double-fluid), slurry initial setting time;

The settlement data set includes: measuring point location, monitoring time, and settlement value.

In the automatic learning step of the settlement prediction model, rolling training is performed on the settlement prediction model according to the new data generated on the construction site, so as to improve the accuracy of the module calculation;

The subsidence prediction model is updated in the background by inputting the shield parameter data newly generated from the last training to the module every K _t time, so as to improve the calculation accuracy of the model.

As an embodiment of the present invention, the element attribute extraction module extracts element attributes; the main factors affecting the surface subsidence are divided into three attribute categories: foundation category, disturbance category and subsidence category;

Each type of meta-attribute includes a variety of excavation characteristics, which are calculated from the original construction data. The specific calculation methods are as follows:

The basic category is the basic information of the tunnel that can be obtained during the construction stage of the shield tunnel, including soil properties, geometric properties and technological properties;

The soil properties specifically include the cohesion, internal friction angle and moisture content of each ring tunnel section. The specific calculation methods of each ring soil property are as follows:

Step S111: Calculate the ratio P _m of each layer of soil in the ring to the excavation surface area respectively, define the buried depth of the upper boundary of the m-th soil layer as d _1,m , the buried depth of the lower boundary as d _2,m , and the buried depth of the center point of the tunnel section. The depth is d, the radius of the excavation surface is R, S _{1, m} is the contact area between the soil body and the excavation surface above the starting elevation of the soil layer, and S _{2, m} is the contact area between the soil body and the excavation surface above the end elevation of the soil layer Area; specific calculation methods include:

Step S112: Calculate the soil properties of the ring tunnel section, define the cohesion, internal friction angle and water content of the m-th soil layer as C _m , φ _m and ω _m , respectively. The friction angle and water content are C, φ and ω, respectively, and the calculation formulas are as follows:

C=∑P _m ·C _m φ=∑P _m ·φ _m ω=∑P _m ·ω _m

The geometric characteristics include the buried depth of each ring of the shield and the soil loss rate in the case of non-grouting. The buried depth of each ring of the shield can be obtained directly from the tunnel design document; the soil loss rate of each ring in the case of non-grouting needs to be obtained. The straight line segment and the curved segment are calculated independently, and the straight line segment only considers the volume of the ring enclosed by the radius R of the shield cutter head and the outer radius r of the tunnel segment. The specific calculation formula is as follows:

v=(R ² -r ² )πl

In order to deflect the shield during the advancement of the curved section, the over-excavation of the soil is usually caused. The theoretical soil loss rate is quite different from that of the straight section. The specific calculation method is as follows:

In the formula, R ₀ tunnel curvature radius, l is the segment width, L is the length of the shield machine, D is the diameter of the shield machine, δ is the over-excavation inside the shield machine, δ′ is the theoretical gap in the middle of the shield machine, and δ′≈δ.

The process characteristics include the type of shield machine, the length of the shield and the diameter of the shield. The types of shield machines are mainly divided into earth pressure balance shield machines and mud-water balance shield machines, which are determined according to the type of shield machine selected for the actual project. ; Shield length and shield diameter are obtained according to the parameter data of the shield machine used in the specific construction.

Disturbance parameters are mainly indicators for evaluating the degree of soil disturbance caused by the shield during the excavation process, which are obtained by analyzing and processing the real-time excavation data of the shield; they are divided into notch disturbance characteristics, attitude disturbance characteristics and shield tail according to the different disturbance mechanisms. perturbation characteristics;

The theoretical pressure difference of shield tunnel is used to evaluate the disturbance of the soil in front of the incision, so as to obtain the disturbance characteristics of the incision. The specific calculation method is as follows:

Step S121, calculating the upper load size Q according to the geological exploration report;

Step S122, calculating the lateral earth pressure coefficient K of the soil body on the excavation surface according to the tunnel geological survey report;

Step S123, extracting the earth pressure data during the advancing period of the ring, and calculating the average earth pressure P of the ring;

Step S124, calculate the theoretical earth pressure difference according to the following formula: ΔP=P-Q·K;

The attitude disturbance characteristics include the attitude change of the shield in the horizontal direction and the attitude change in the elevation direction of the shield; the attitude change in the elevation direction of the shield is represented by the change in the pitch angle of the shield, namely:

为盾构环俯仰角变化量，

为每环开始时的俯仰角，

为每环结束时的俯仰角；in the formula

is the variation of the pitch angle of the shield ring,

is the pitch angle at the beginning of each loop,

is the pitch angle at the end of each loop;

The attitude change of the shield in the horizontal direction is represented by the change of the horizontal angle, which is calculated according to the deviation of the shield and the linearity of the axis. The specific calculation method is as follows:

In the formula, h′ _x is the horizontal deviation of the incision at the beginning of each ring, h″ _x is the horizontal deviation of the incision at the end of each ring, t′ _x is the horizontal deviation of the shield tail at the beginning of each ring, and t″ _x is the horizontal deviation of each ring. The amount of horizontal deviation of the shield tail at the end of the ring;

The grouting pressure of the shield tail is obtained by calculating the average pressure of the shield tail grouting in each ring, and the filling rate of the shield tail grouting is calculated according to the ratio of the cumulative grouting amount of the ring to the theoretical shield tail gap, so as to obtain the shield tail disturbance characteristics. The formula is as follows:

In the formula, g is the grouting filling rate of the shield tail, Gt is the cumulative grouting amount of the shield tail ring, and υ is the theoretical shield tail gap size;

The subsidence category mainly describes the subsidence in the process of shield tunneling through the subsidence form and subsidence amplitude;

The settlement form is mainly represented by the width coefficient of the lateral settlement tank. The width coefficient I of the lateral settlement tank is obtained by using the peck formula to fit the curve of the lateral settlement measuring points within the influence range and take the average value. The peck formula is as follows:

where S is the settlement value of any point on the surface; S _max is the maximum surface settlement, located just above the tunnel axis; x is the horizontal distance between any point and the tunnel axis; I is the settlement tank width coefficient, that is, the tunnel axis and the settlement tank the distance of the inflection point;

The subsidence range includes the average subsidence 10m in front of the incision, the average subsidence above the shield body, and the average subsidence 10m behind the shield tail.

As an embodiment of the present invention, in the training step of the settlement data generator, after extracting the meta-attributes from the construction data of different projects obtained according to the above method, the basic and disturbance attributes are used as model inputs, and the settlement attributes are used as model inputs. As the model output, a recurrent neural network model is trained as a settlement data generator; the specific steps are as follows:

Step S21, data preprocessing, convert the meta-attribute data into time series data, and divide the training set and the test set;

Step S22, constructing a three-layer cyclic neural network model;

Step S23, using the training set data for model training, and verifying the model prediction effect on the test set, and adjusting the model parameters according to the results;

Step S24, save the model.

In the step of generating subsidence data, basic data are calculated according to the geometric data, geological data and shield machine parameter data of the actual application project, and then the disturbance data range is calculated in combination with the theoretical calculation method.

When calculating the notch disturbance characteristics, the earth pressure fluctuation in the actual propulsion process is simulated according to the theoretical earth pressure value plus a certain random fluctuation amount, and then the simulated theoretical earth pressure difference is calculated; when calculating the shield tail disturbance characteristics, the average grouting The pressure is simulated according to the grouting pressure range under similar soil layers, and the average grouting rate fluctuates up and down according to the theoretical grouting filling rate of 150%; when calculating the attitude disturbance characteristics, the theoretical shield tunnel is linearly obtained according to the design axis of the shield tunnel Elevation and horizontal attitude changes.

Input the simulated construction data generated by the above method into the trained settlement data generator to obtain the settlement attributes at each moment. According to the calculated settlement properties, the settlement amount of each point is obtained with the help of the empirical settlement curve formula. The specific steps include:

Step S31, obtaining the actual construction project tunnel geometric data, geological data and shield machine parameter data;

Step S32, designing a simulated construction scheme according to the basic data;

Step S33, input the simulated construction data into the settlement data generator to obtain settlement data;

Step S34: Calculate the settlement of each settlement monitoring point at each moment according to the settlement data, and the specific calculation method is as follows:

Firstly, according to the calculated settlement data combined with the empirical longitudinal settlement curve formula, the longitudinal settlement curve is obtained. The empirical formula is as follows:

In the formula, S _y is the settlement of the settlement monitoring point at the distance x from the incision on the axis, y is the distance between the monitoring point and the incision, and y is a negative number when the monitoring point is in front of the incision;

Then combined with the lateral settlement tank coefficient I, the lateral settlement tank formula of the settlement monitoring point can be calculated:

Finally, by combining the above two formulas, the settlement value of the settlement monitoring point whose relative incision coordinates are (x, y) can be calculated.

Step S35, save the settlement simulation data.

As an embodiment of the present invention, in the pre-training step of the settlement prediction model, the LSTM monitoring point settlement prediction model is constructed according to the shield construction simulation data calculated by the settlement data generator. Soil cohesion on the excavation face, friction angle in the soil on the excavation face, moisture content of the soil on the excavation face, upper load, frontal earth pressure, grouting filling rate, grouting pressure, change in the horizontal attitude of the shield, shield The attitude change in the elevation direction, the longitudinal distance between the settlement monitoring point and the incision, and the lateral distance between the settlement monitoring point and the axis; the model output is the cumulative settlement value at time t; among them, the single sample time step n of the LSTM model is 12, and the LSTM layer The number is 1.

The real-time settlement prediction module uses a settlement prediction model to predict the settlement of the settlement monitoring points within the influence range of the shield, which specifically includes the following steps:

Step S51, obtaining the current shield position, and determining a set of monitoring points for which settlement prediction needs to be performed, specifically including horizontal and vertical monitoring points within the range of ten meters in front of the incision to ten meters behind the shield tail;

Step S52, obtaining historical shield construction data from time t-n to time t;

Step S53, obtaining the setting values of the shield construction parameters of the next 5 time steps;

Step S54, input the shield construction data from time t-n+i to time t+i, and use the LSTM model to predict the settlement of each monitoring point at time t+i;

Step S55, repeating step S54 until the settlement prediction for the next five time steps is completed.

The beneficial effect of the present invention is that: the system and method for predicting the ground settlement of shield construction proposed by the present invention can improve the applicability and accuracy of the prediction, and does not require a large amount of data accumulation in the early stage.

The invention constructs a settlement data generator through a large amount of shield construction project data, masters the general laws of settlement changes in the shield tunneling process under different environments, and quickly establishes an accurate settlement prediction model in combination with the actual construction data of the project, and solves the problem of existing data-driven prediction. The method has poor adaptability and requires a large amount of data accumulation in the early stage.

Description of drawings

FIG. 1 is a schematic diagram of the composition of a surface settlement prediction system for shield construction in an embodiment of the present invention.

FIG. 2 is a schematic diagram of three attribute categories extracted by a meta-attribute extraction module according to an embodiment of the present invention.

FIG. 3 is a schematic diagram of calculating the theoretical soil loss volume in an embodiment of the present invention.

Detailed ways

The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

In order to further understand the present invention, the preferred embodiments of the present invention are described below in conjunction with the examples, but it should be understood that these descriptions are only for further illustrating the features and advantages of the present invention, rather than limiting the claims of the present invention.

The description in this section is only for a few typical embodiments, and the present invention is not limited to the scope of the description of the embodiments. It is also within the scope of the description and protection of the present invention to replace some technical features in the embodiments with the same or similar prior art means.

The descriptions of the steps in the various embodiments in the specification are only for the convenience of description, and the implementation manner of the present application is not limited by the order in which the steps are implemented. The "connection" in the specification includes both direct connection and indirect connection.

The present invention discloses a surface settlement prediction system for shield construction. FIG. 1 is a schematic diagram of the composition of the shield construction surface settlement prediction system in an embodiment of the present invention; please refer to FIG. 1 , the shield construction surface settlement prediction system includes: Meta-attribute extraction module 1 , settlement data generator training module 2 , settlement data generation module 3 , settlement prediction model pre-training module 4 and real-time settlement prediction module 5 .

The meta-attribute extraction module 1 is used to recombine the original data set to extract meta-attributes, and calculate each attribute characteristic index; the subsidence data generator training module 2 is used to construct a meta-attribute-based subsidence data generator; The settlement data generation module 3 is used to generate a set of simulation data in combination with the engineering characteristics of the current construction project; the settlement prediction model pre-training module 4 is used to train the settlement prediction model of the current project in combination with the generated simulation data to obtain the initial settlement prediction model; The real-time settlement prediction module 5 is used to obtain real-time shield tunneling data to predict the settlement value of the surface settlement monitoring point.

Please continue to refer to FIG. 1 , in an embodiment of the present invention, the system further includes: a historical construction project data acquisition module 6 , and a settlement prediction model automatic learning module 7 . The historical construction project data acquisition module 6 is used to collect shield construction data, excavation parameter data, geological parameter data and land subsidence data under different construction environments; the settlement prediction model automatic learning module 7 is used to update the settlement prediction model online.

In an embodiment of the present invention, the meta-attribute extraction module 1 is used to extract element attributes; the main factors affecting the surface subsidence are divided into three attribute categories: foundation, disturbance and settlement. FIG. 2 is a schematic diagram of three attribute categories extracted by a meta-attribute extraction module in an embodiment of the present invention; please refer to FIG. 2 , which discloses three attribute categories extracted by the meta-attribute extraction module.

In the process of shield tunneling, the surface subsidence is mainly caused by the change of the stratum stress field caused by the cutting of the cutter head, the friction between the shield machine and the surrounding soil layer, and the gap formed between the shield tail and the surrounding soil layer. Due to the difference in surface geological conditions, the reflection form of surface subsidence is also different. Based on the above analysis of the mechanism of surface subsidence during shield tunneling, the main factors affecting surface subsidence are divided into three attributes: foundation, disturbance and subsidence. category. Each type of meta-attribute includes a variety of excavation characteristics, which are calculated from raw construction data.

The basic class is the basic information required for the construction of shield tunnels, including soil properties, geometric properties and technological properties. The soil properties specifically include the cohesion, internal friction angle and moisture content of each ring tunnel section. The specific calculation methods for each ring soil properties include:

Step S111: Calculate the ratio P _m of each layer of soil in the ring to the excavation surface area respectively, define the buried depth of the upper boundary of the m-th soil layer as d _1,m , the buried depth of the lower boundary as d _2,m , and the buried depth of the center point of the tunnel section. The depth is d, the radius of the excavation surface is R, S _{1, m} is the contact area between the soil body and the excavation surface above the starting elevation of the soil layer, and S _{2, m} is the contact area between the soil body and the excavation surface above the end elevation of the soil layer Area; specific calculation methods include:

Step S112: Calculate the soil properties of the ring tunnel section, define the cohesion, internal friction angle and water content of the m-th soil layer as C _m , φ _m and ω _m , respectively. The friction angle and water content are C, φ and ω, respectively, and the calculation formulas are as follows:

C=∑P _m ·C _m φ=∑P _m ·φ _m ω=∑P _m ·ω _m

The geometric characteristics include the buried depth of each ring of the shield and the soil loss rate in the case of non-grouting. The buried depth of each ring of the shield can be obtained directly from the tunnel design document; the soil loss rate of each ring in the case of non-grouting needs to be obtained. The straight line segment and the curved segment are calculated independently, and the straight line segment only considers the volume of the ring enclosed by the radius R of the shield cutter head and the outer radius r of the tunnel segment. The specific calculation formula is as follows:

v=(R ² -r ² )πl

In order to deflect the shield during the advancement of the curved section, the over-excavation of the soil is usually caused, and the theoretical soil loss rate is quite different from that of the straight section (see Figure 3). The specific calculation method is as follows:

In the formula, R ₀ tunnel curvature radius, l is the segment width, L is the length of the shield machine, D is the diameter of the shield machine, δ is the over-excavation inside the shield machine, δ′ is the theoretical gap in the middle of the shield machine, and δ′≈δ.

The process characteristics include the type of shield machine, the length of the shield and the diameter of the shield. The types of shield machines are mainly divided into earth pressure balance shield machines and mud-water balance shield machines, which are determined according to the type of shield machine selected for the actual project. ; Shield length and shield diameter are obtained according to the parameter data of the shield machine used in the specific construction.

Disturbance parameters are mainly indicators for evaluating the degree of soil disturbance caused by the shield during the excavation process, which are obtained by analyzing and processing the real-time excavation data of the shield; they are divided into notch disturbance characteristics, attitude disturbance characteristics and shield tail according to the different disturbance mechanisms. perturbation characteristics.

The soil in front of the incision is mainly affected by the pressure change of the soil bin before the shield machine arrives. The soil in front of the excavation is squeezed or unloaded, causing the ground to bulge or settle; the traditional soil bin pressure setting method is based on The theoretical earth pressure in front of the incision is calculated, and the earth pressure is fine-tuned according to the actual settlement in front of the incision. Generally, the pressure of the soil bin is slightly larger than the theoretical earth pressure. In order to comprehensively consider the above factors, the theoretical pressure difference of shield tunnel is used to evaluate the disturbance of the soil in front of the incision, so as to obtain the disturbance characteristics of the incision. The specific calculation methods include:

Step S121, calculating the upper load size Q according to the geological exploration report;

Step S122, calculating the lateral earth pressure coefficient K of the soil body on the excavation surface according to the tunnel geological survey report;

Step S123, extracting the earth pressure data during the advancing period of the ring, and calculating the average earth pressure P of the ring;

Step S124, calculate the theoretical earth pressure difference according to the following formula: ΔP=P-Q·K;

Since the adjustment of the attitude of the shield and the friction between the shield shell and the surrounding soil will disturb the soil around the shield and cause the soil above the shield to settle, and the adjustment of the attitude of the shield has a significant impact on the settlement, the shield is adopted. The attitude adjustment range is used to measure the disturbance degree of the surrounding soil due to the change of the shield attitude.

The attitude disturbance characteristics include the attitude change of the shield in the horizontal direction and the attitude change in the elevation direction of the shield; the attitude change in the elevation direction of the shield is represented by the change in the pitch angle of the shield, namely:

为盾构环俯仰角变化量，

为每环开始时的俯仰角，

为每环结束时的俯仰角。In the formula,

is the variation of the pitch angle of the shield ring,

is the pitch angle at the beginning of each loop,

is the pitch angle at the end of each loop.

The attitude change of the shield in the horizontal direction is represented by the change of the horizontal angle, which is calculated according to the deviation of the shield and the linearity of the axis. The specific calculation method is as follows:

In the formula, h' _x is the horizontal deviation of the incision at the beginning of each ring, h" _x is the horizontal deviation of the incision at the end of each ring, t' _x is the horizontal deviation of the shield tail at the beginning of each ring, and t" _x is the horizontal deviation of each ring. The amount of horizontal offset of the shield tail at the end of the ring.

The main factors that cause the soil behind the shield tail are the gap generated after the segment comes out of the shield tail, the grouting amount of the shield tail and the grouting pressure value. This patent uses the shield tail grouting filling rate and the shield tail grouting pressure to reflect the shield tail. soil disturbance.

The grouting pressure of the shield tail is obtained by calculating the average pressure of the shield tail grouting in each ring, and the filling rate of the shield tail grouting is calculated according to the ratio of the cumulative grouting amount of the ring to the theoretical shield tail gap, so as to obtain the shield tail disturbance characteristics. The formula is as follows:

In the formula, g is the filling rate of the shield tail grouting, G _t is the cumulative grouting amount of the shield tail ring, and v is the theoretical shield tail gap size.

The subsidence category mainly describes the subsidence size in the process of shield tunneling through the subsidence form and subsidence amplitude. The settlement form is mainly represented by the width coefficient of the lateral settlement tank. The width coefficient I of the lateral settlement tank is obtained by using the peck formula to fit the curve of the lateral settlement measuring points within the influence range and take the average value. The peck formula is as follows:

In the formula, S is the settlement value of any point on the surface; S _max is the maximum surface settlement, located directly above the tunnel axis; x is the horizontal distance between any point and the tunnel axis; I is the width coefficient of the settlement tank, that is, the tunnel axis and the settlement The distance from the inflection point of the slot.

The subsidence range includes the average subsidence 10m in front of the incision, the average subsidence above the shield body, and the average subsidence 10m behind the shield tail.

In an embodiment of the present invention, the subsidence data generator training module 2 is used to extract the meta-attributes from the acquired construction data of different projects according to the above method, and then use the basic class and disturbance class attributes as model input, and the settlement class The attribute is used as the model output, and the recurrent neural network model is trained as the sedimentation data generator; the specific steps include:

Step S21, data preprocessing, convert the meta-attribute data into time series data, and divide the training set and the test set;

Step S22, constructing a three-layer cyclic neural network model;

Step S23, using the training set data for model training, and verifying the model prediction effect on the test set, and adjusting the model parameters according to the results;

Step S24, save the model.

The subsidence data generation module 3 is used for calculating basic data according to the geometric data, geological data and shield machine parameter data of the actual application project, and then calculating the disturbance data range in combination with the theoretical calculation method.

When calculating the notch disturbance characteristics, the earth pressure fluctuation in the actual propulsion process is simulated according to the theoretical earth pressure value plus a certain random fluctuation amount, and then the simulated theoretical earth pressure difference is calculated; when calculating the shield tail disturbance characteristics, the average grouting The pressure is simulated according to the grouting pressure range under similar soil layers, and the average grouting rate fluctuates up and down according to the theoretical grouting filling rate of 150%; when calculating the attitude disturbance characteristics, the theoretical shield tunnel is linearly obtained according to the design axis of the shield tunnel Elevation and horizontal attitude changes.

Input the simulated construction data generated by the above method into the trained settlement data generator to obtain the settlement attributes at each moment; obtain the settlement amount of each point with the help of the empirical settlement curve formula according to the calculated settlement attributes. The specific steps are: include:

Step S31, obtaining the actual construction project tunnel geometric data, geological data and shield machine parameter data;

Step S32, designing a simulated construction scheme according to the basic data;

Step S33, input the simulated construction data into the settlement data generator to obtain settlement data;

Step S34: Calculate the settlement of each settlement monitoring point at each moment according to the settlement data, and the specific calculation method is as follows:

Firstly, according to the calculated settlement data combined with the empirical longitudinal settlement curve formula, the longitudinal settlement curve is obtained. The empirical formula is as follows:

In the formula, S _y is the settlement of the settlement monitoring point at the distance x from the incision on the axis, y is the distance between the monitoring point and the incision, and y is a negative number when the monitoring point is in front of the incision;

Then combined with the lateral settlement tank coefficient I, the lateral settlement tank formula of the settlement monitoring point can be calculated:

Finally, by combining the above two formulas, the settlement value of the settlement monitoring point whose relative incision coordinates are (x, y) can be calculated.

Step S35, save the settlement simulation data.

In an embodiment of the present invention, the settlement prediction model pre-training module 4 is used to construct an LSTM monitoring point settlement prediction model according to the shield construction simulation data calculated by the settlement data generator, and the input of the LSTM model is t-n to t. Soil cohesion on the excavation surface, friction angle in the soil body on the excavation surface, moisture content of the soil on the excavation surface, upper load, frontal earth pressure, grouting filling rate, grouting pressure, the amount of change in the horizontal direction of the shield, shield The attitude change in the structural elevation direction, the longitudinal distance between the settlement monitoring point and the incision, and the lateral distance between the settlement monitoring point and the axis; the model output is the cumulative settlement value at time t; among them, the single sample time step n of the LSTM model is 12, LSTM The number of layers is 1;

The real-time settlement prediction module 5 is used to predict the settlement of the settlement monitoring point within the influence range of the shield by adopting the settlement prediction model, and specifically includes the following steps:

Step S51, obtaining the current shield position, and determining a set of monitoring points for which settlement prediction needs to be performed, specifically including horizontal and vertical monitoring points within the range of ten meters in front of the incision to ten meters behind the shield tail;

Step S52, obtaining historical shield construction data from time t-n to time t;

Step S53, obtaining the setting values of the shield construction parameters of the next 5 time steps;

Step S54, input the shield construction data from time t-n+i to time t+i, and use the LSTM model to predict the settlement of each monitoring point at time t+i;

Step S55, repeating step S54 until the settlement prediction for the next five time steps is completed.

The historical construction project data acquisition module 6 is used to obtain the shield construction of different engineering projects from the shield construction database (through the shield construction database, historical construction data of other construction projects can be acquired, which can be used for model pre-training). data, and classify the obtained original construction data, extract the data related to settlement prediction, and generate the original data set; the original data set specifically includes geological data set, tunnel geometry data set, shield machine parameter data set, shield Excavation data set and settlement data set, the data contained in each type of data set are as follows:

The geological data set includes: drilling point position, starting and ending depth of each soil layer, soil mechanical parameters of each soil layer; soil mechanical parameters of each soil layer include cohesion, internal friction angle, compressibility, compressive modulus, natural pores ratio, lateral earth pressure coefficient, natural moisture content, severity;

The tunnel geometry dataset includes: the outer diameter of the tunnel segment, the buried depth of the tunnel, and the line type of the tunnel;

Shield machine parameter data set includes: shield diameter, shield length, shield process type;

The shield tunneling data set includes: shield tunneling process, cut horizontal deviation, cut elevation deviation, shield tail horizontal deviation, shield tail elevation deviation, shield slope angle, frontal earth pressure (zone), grouting amount (distribution), Slurry pressure (distribution), tunneling speed, total thrust, slurry type (single-fluid/double-fluid), slurry initial setting time;

The settlement data set includes: measuring point location, monitoring time, and settlement value.

The automatic learning module 7 of the settlement prediction model is used to perform rolling training on the settlement prediction model according to the new data generated on the construction site, so as to improve the accuracy of the calculation of the module; through every _Kt time interval, input new data from the last training to this module. The generated shield parameter data is used to update the settlement prediction model in the background to improve the calculation accuracy of the model.

The present invention also discloses a method for predicting surface settlement of shield construction, the method for predicting surface settlement of shield construction includes:

[Step S0] The step of obtaining historical construction project data, collecting shield construction data, excavation parameter data, geological parameter data and land subsidence data under different construction environments;

[Step S1] Meta-attribute extraction step, the original data set is recombined to extract the meta-attributes, and each attribute characteristic index is calculated;

[Step S2] the training step of the settlement data generator, constructing a settlement data generator based on meta-attributes;

[Step S3] the step of generating settlement data, generating a set of simulation data in combination with the engineering characteristics of the current construction project;

[Step S4] the pre-training step of the settlement prediction model, combining the generated simulation data to train the settlement prediction model of the current project to obtain the initial settlement prediction model;

[Step S5] The step of real-time settlement prediction is to obtain real-time shield tunneling data to predict the settlement value of the surface settlement monitoring point.

[Step S6 ] The automatic learning step of the settlement prediction model is to update the settlement prediction model online.

In one embodiment, the method of the present invention may not include part of the above steps, for example, may not include steps S0 and S6.

In the historical construction project data acquisition step, the shield construction data of different engineering projects is acquired from the shield construction database, the acquired original construction data is classified, the data related to settlement prediction is extracted, and an original data set is generated. ; The original data set specifically includes geological data set, tunnel geometry data set, shield machine parameter data set, shield excavation data set and subsidence data set. The data contained in each type of data set are as follows:

The geological data set includes: drilling point position, starting and ending depth of each soil layer, soil mechanical parameters of each soil layer; soil mechanical parameters of each soil layer include cohesion, internal friction angle, compressibility, compressive modulus, natural pores ratio, lateral earth pressure coefficient, natural moisture content, severity;

The tunnel geometry dataset includes: the outer diameter of the tunnel segment, the buried depth of the tunnel, and the line type of the tunnel;

Shield machine parameter data set includes: shield diameter, shield length, shield process type;

The shield tunneling data set includes: shield tunneling process, cut horizontal deviation, cut elevation deviation, shield tail horizontal deviation, shield tail elevation deviation, shield slope angle, frontal earth pressure (zone), grouting amount (distribution), Slurry pressure (distribution), tunneling speed, total thrust, slurry type (single-fluid/double-fluid), slurry initial setting time;

The settlement data set includes: measuring point location, monitoring time, and settlement value.

In an embodiment of the present invention, the element attribute extraction module extracts element attributes, and divides the main factors affecting the surface subsidence into three attribute categories: foundation, disturbance and settlement.

In the process of shield tunneling, the surface subsidence is mainly caused by the change of the stratum stress field caused by the cutting of the cutter head, the friction between the shield machine and the surrounding soil layer, and the gap formed between the shield tail and the surrounding soil layer. Due to the differences in the surface geological conditions, the reflected forms of surface subsidence are also different. Based on the above analysis of the mechanism of the surface subsidence caused by the shield tunneling process, this patent divides the main factors affecting the surface subsidence into three attribute categories: foundation, disturbance and subsidence.

Each type of meta-attribute includes a variety of excavation characteristics, which are calculated from raw construction data.

The foundation category is the basic information of the tunnel that can be obtained during the construction stage of the shield tunnel, including soil properties, geometric properties and technological properties; the soil properties include the cohesion, internal friction angle and moisture content of each ring tunnel section, The specific calculation methods of characteristics include:

Step S111: Calculate the ratio P _m of each layer of soil in the ring to the excavation surface area respectively, define the buried depth of the upper boundary of the m-th soil layer as d _1,m , the buried depth of the lower boundary as d _2,m , and the buried depth of the center point of the tunnel section. The depth is d, the radius of the excavation surface is R, S _{1, m} is the contact area between the soil body and the excavation surface above the starting elevation of the soil layer, and S _{2, m} is the contact area between the soil body and the excavation surface above the end elevation of the soil layer Area; specific calculation methods include:

Step S112: Calculate the soil properties of the ring tunnel section, define the cohesion, internal friction angle and water content of the m-th soil layer as C _m , φ _m and ω _m , respectively. The friction angle and water content are C, φ and ω, respectively, and the calculation formulas are as follows:

C=∑P _m ·C _m φ=∑P _m ·φ _m ω=∑P _m ·ω _m

The geometric characteristics include the buried depth of each ring of the shield and the soil loss rate in the case of non-grouting. The buried depth of each ring of the shield can be obtained directly from the tunnel design document; the soil loss rate of each ring in the case of non-grouting needs to be obtained. The straight line segment and the curved segment are calculated independently, and the straight line segment only considers the volume of the ring enclosed by the radius R of the shield cutter head and the outer radius r of the tunnel segment. The specific calculation formula is as follows:

v=(R ² -r ² )πl

In order to deflect the shield during the advancement of the curved section, the over-excavation of the soil is usually caused. The theoretical soil loss rate is quite different from that of the straight section. The specific calculation method is as follows:

In the formula, R ₀ tunnel curvature radius, l is the segment width, L is the length of the shield machine, D is the diameter of the shield machine, δ is the over-excavation inside the shield machine, δ′ is the theoretical gap in the middle of the shield machine, and δ′≈δ.

The process characteristics include the type of shield machine, the length of the shield and the diameter of the shield. The types of shield machines are mainly divided into earth pressure balance shield machines and mud-water balance shield machines, which are determined according to the type of shield machine selected for the actual project. ; Shield length and shield diameter are obtained according to the parameter data of the shield machine used in the specific construction.

Disturbance parameters are mainly indicators for evaluating the degree of soil disturbance caused by the shield during the excavation process, which are obtained by analyzing and processing the real-time excavation data of the shield; they are divided into notch disturbance characteristics, attitude disturbance characteristics and shield tail according to the different disturbance mechanisms. perturbation characteristics.

The soil in front of the incision is mainly affected by the pressure change of the soil bin before the shield machine arrives. The soil in front of the excavation is squeezed or unloaded, causing the ground to bulge or settle; the traditional soil bin pressure setting method is based on The theoretical earth pressure in front of the incision is calculated, and the earth pressure is fine-tuned according to the actual settlement in front of the incision. Generally, the pressure of the soil bin is slightly larger than the theoretical earth pressure. In order to comprehensively consider the above factors, the theoretical pressure difference of shield tunnel is used to evaluate the disturbance of the soil in front of the incision, so as to obtain the disturbance characteristics of the incision. The specific calculation methods include:

Step S121, calculating the upper load size Q according to the geological exploration report;

Step S122, calculating the lateral earth pressure coefficient K of the soil body on the excavation surface according to the tunnel geological survey report;

Step S123, extracting the earth pressure data during the advancing period of the ring, and calculating the average earth pressure P of the ring;

Step S124, calculate the theoretical earth pressure difference according to the following formula: ΔP=P-Q·K.

Since the adjustment of the attitude of the shield and the friction between the shield shell and the surrounding soil will disturb the soil around the shield and cause the soil above the shield to settle, and the adjustment of the attitude of the shield has a significant impact on the settlement, the shield is adopted. The attitude adjustment range is used to measure the disturbance degree of the surrounding soil due to the change of the shield attitude.

The attitude disturbance characteristics include the attitude change of the shield in the horizontal direction and the attitude change in the elevation direction of the shield; the attitude change in the elevation direction of the shield is represented by the change in the pitch angle of the shield, namely:

为盾构环俯仰角变化量，

为每环开始时的俯仰角，

为每环结束时的俯仰角；in the formula

is the variation of the pitch angle of the shield ring,

is the pitch angle at the beginning of each loop,

is the pitch angle at the end of each loop;

The attitude change of the shield in the horizontal direction is represented by the change of the horizontal angle, which is calculated according to the deviation of the shield and the linearity of the axis. The specific calculation method is as follows:

In the formula, h' _x is the horizontal deviation of the incision at the beginning of each ring, h" _x is the horizontal deviation of the incision at the end of each ring, t' _x is the horizontal deviation of the shield tail at the beginning of each ring, and t" _x is the horizontal deviation of each ring. The amount of horizontal offset of the shield tail at the end of the ring.

The main factors that cause the soil behind the shield tail are the gap generated after the segment comes out of the shield tail, the grouting amount of the shield tail and the grouting pressure value. This patent uses the shield tail grouting filling rate and the shield tail grouting pressure to reflect the shield tail. soil disturbance. The grouting pressure of the shield tail is obtained by calculating the average pressure of the shield tail grouting in each ring, and the filling rate of the shield tail grouting is calculated according to the ratio of the cumulative grouting amount of the ring to the theoretical shield tail gap, so as to obtain the shield tail disturbance characteristics. The formula is as follows:

In the formula, g is the filling rate of the shield tail grouting, G _t is the cumulative grouting amount of the shield tail ring, and v is the theoretical shield tail gap size.

The settlement category mainly describes the settlement size during the shield propulsion process by the settlement form and the settlement amplitude; the settlement form is mainly expressed by the width coefficient of the lateral settlement tank, and the width coefficient I of the lateral settlement tank is used to measure the lateral settlement within the influence range by using the peck formula. The points are curve-fitted and averaged, and the peck formula is as follows:

In the formula, S is the settlement value of any point on the surface; S _max is the maximum surface settlement, located directly above the tunnel axis; x is the horizontal distance between any point and the tunnel axis; I is the width coefficient of the settlement tank, that is, the tunnel axis and the settlement The distance from the inflection point of the slot. The subsidence range includes the average subsidence 10m in front of the incision, the average subsidence above the shield body, and the average subsidence 10m behind the shield tail.

In an embodiment of the present invention, in the training step of the settlement data generator, after extracting the meta-attributes from the acquired construction data of different projects according to the above method, the foundation class and disturbance class attributes are used as model input, and the settlement class attributes are used as model inputs. As the model output, a recurrent neural network model is trained as a settlement data generator; the specific steps are as follows:

Step S21, data preprocessing, convert the meta-attribute data into time series data, and divide the training set and the test set;

Step S22, constructing a three-layer cyclic neural network model;

Step S23, using the training set data for model training, and verifying the model prediction effect on the test set, and adjusting the model parameters according to the results;

Step S24, save the model.

In the step of generating subsidence data, basic data are calculated according to the geometric data, geological data and shield machine parameter data of the actual application project, and then the disturbance data range is calculated in combination with the theoretical calculation method.

When calculating the notch disturbance characteristics, the earth pressure fluctuation in the actual propulsion process is simulated according to the theoretical earth pressure value plus a certain random fluctuation amount, and then the simulated theoretical earth pressure difference is calculated; when calculating the shield tail disturbance characteristics, the average grouting The pressure is simulated according to the grouting pressure range under similar soil layers, and the average grouting rate fluctuates up and down according to the theoretical grouting filling rate of 150%; when calculating the attitude disturbance characteristics, the theoretical shield tunnel is linearly obtained according to the design axis of the shield tunnel Elevation and horizontal attitude changes.

Input the simulated construction data generated by the above method into the trained settlement data generator to obtain the settlement attributes at each moment. According to the calculated settlement properties, the settlement of each point can be obtained by means of the empirical settlement curve formula. The specific steps are as follows:

Step S31, obtaining the actual construction project tunnel geometric data, geological data and shield machine parameter data;

Step S32, designing a simulated construction scheme according to the basic data;

Step S33, input the simulated construction data into the settlement data generator to obtain settlement data;

Step S34: Calculate the settlement of each settlement monitoring point at each moment according to the settlement data, and the specific calculation method is as follows:

Firstly, according to the calculated settlement data combined with the empirical longitudinal settlement curve formula, the longitudinal settlement curve is obtained. The empirical formula is as follows:

In the formula, S _y is the settlement of the settlement monitoring point at the distance x from the incision on the axis, y is the distance between the monitoring point and the incision, and y is a negative number when the monitoring point is in front of the incision;

Then combined with the lateral settlement tank coefficient I, the lateral settlement tank formula of the settlement monitoring point can be calculated:

Finally, by combining the above two formulas, the settlement value of the settlement monitoring point whose relative incision coordinates are (x, y) can be calculated.

Step S35, save the settlement simulation data.

In an embodiment of the present invention, in the pre-training step of the settlement prediction model, the LSTM monitoring point settlement prediction model is constructed according to the shield construction simulation data calculated by the settlement data generator, and the input of the LSTM model is the opening time from t-n to t. Soil cohesion on the excavation face, friction angle in the soil on the excavation face, moisture content of the soil on the excavation face, upper load, frontal earth pressure, grouting filling rate, grouting pressure, change in the horizontal attitude of the shield, shield The attitude change in the elevation direction, the longitudinal distance between the settlement monitoring point and the incision, and the lateral distance between the settlement monitoring point and the axis; the model output is the cumulative settlement value at time t; among them, the single sample time step n of the LSTM model is 12, and the LSTM layer The number is 1.

In an embodiment of the present invention, in the real-time settlement prediction step, a settlement prediction model is used to predict the settlement of the settlement monitoring points within the influence range of the shield, which specifically includes the following steps:

Step S51, obtaining the current shield position, and determining a set of monitoring points for which settlement prediction needs to be performed, specifically including horizontal and vertical monitoring points within the range of ten meters in front of the incision to ten meters behind the shield tail;

Step S52, obtaining historical shield construction data from time t-n to time t;

Step S53, obtaining the setting values of the shield construction parameters of the next 5 time steps;

Step S54, input the shield construction data from time t-n+i to time t+i, and use the LSTM model to predict the settlement of each monitoring point at time t+i;

Step S55, repeating step S54 until the settlement prediction for the next five time steps is completed.

In the automatic learning step of the settlement prediction model, rolling training is carried out on the settlement prediction model according to the new data generated on the construction site, so as to improve the accuracy of the calculation of the module; through every _Kt time interval, the module is input to the module which is newly generated from the last training. Based on the shield parameter data, the subsidence prediction model is updated in the background to improve the calculation accuracy of the model.

To sum up, the system and method for predicting the surface settlement of shield tunnel construction proposed by the present invention can improve the applicability and accuracy of the prediction, and does not require a large amount of data accumulation in the early stage.

The invention constructs a settlement data generator through a large amount of shield construction project data, grasps the general laws of settlement changes in the shield tunneling process under different environments, and quickly establishes an accurate settlement prediction model in combination with the actual construction data of the project, and solves the problem of existing data-driven prediction. The method has poor adaptability and requires a large amount of data accumulation in the early stage.

It should be noted that the present application may be implemented in software and/or a combination of software and hardware; for example, may be implemented using an application specific integrated circuit (ASIC), a general purpose computer, or any other similar hardware device. In some embodiments, the software program of the present application may be executed by a processor to implement the above steps or functions. Likewise, the software programs of the present application (including associated data structures) may be stored on a computer-readable recording medium; for example, RAM memory, magnetic or optical drives or floppy disks, and the like. In addition, some steps or functions of the present application may be implemented in hardware; for example, as a circuit that cooperates with a processor to perform various steps or functions.

The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity, all possible combinations of the technical features in the above-described embodiments are not described. However, as long as there is no contradiction between the combinations of these technical features, All should be regarded as the scope described in this specification.

The description and application of the present invention herein is illustrative, and is not intended to limit the scope of the present invention to the above-described embodiments. The effects or advantages involved in the embodiments may be interfered by various factors and may not be embodied in the embodiments, and the description of the effects or advantages is not intended to limit the embodiments. Variations and variations of the embodiments disclosed herein are possible, and alternative and equivalent various components of the embodiments are known to those of ordinary skill in the art. It should be apparent to those skilled in the art that the present invention may be implemented in other forms, structures, arrangements, proportions, and with other components, materials and components without departing from the spirit or essential characteristics of the invention. Other modifications and changes of the embodiments disclosed herein may be made without departing from the scope and spirit of the invention.

Claims (4)

1. The utility model provides a shield constructs construction earth's surface settlement prediction system which characterized in that, shield constructs construction earth's surface settlement prediction system includes:

the meta-attribute extraction module is used for recombining the original data sets to extract meta-attributes and calculating each attribute characteristic index;

the settlement data generator training module is used for constructing a settlement data generator based on the meta-attribute;

the settlement data generation module is used for generating a group of simulation data by combining the engineering characteristics of the current construction project;

the settlement prediction model pre-training module is used for training a settlement prediction model of the current engineering by combining the generated simulation data to obtain an initial settlement prediction model;

the real-time settlement prediction module is used for obtaining real-time shield tunneling data and predicting settlement values of the ground surface settlement monitoring points;

the element attribute extraction module is used for extracting element attributes; dividing main factors influencing surface subsidence into three attribute categories of a basic category, a disturbance category and a subsidence category;

each type of element attribute comprises a plurality of tunneling characteristics which are obtained by calculating original construction data, and the specific calculation method comprises the following steps:

the foundation type is foundation information required by the construction of the shield tunnel, and comprises soil property, geometric property and process property;

the soil property characteristics specifically comprise cohesive force, an internal friction angle and water content of each ring of tunnel section, and the specific calculation mode of each ring of soil property characteristics comprises the following steps:

step S111, calculating the area proportion P of each layer of soil of the ring to the excavation surface respectively_mDefining the burial depth of the boundary on the mth layer of soil layer as d_1，mLower boundary buried depth is d_2，mThe buried depth of the center point of the tunnel section is d, and the radius of the excavation surface is R and S_1，mThe contact area S of the soil body above the initial elevation of the soil layer and the excavation surface_2，mThe contact area between the soil body and the excavation surface is larger than the finished elevation of the soil layer; the specific calculation method comprises the following steps:

step S112, calculating the soil property of the section of the circular tunnel, and defining the cohesive force, the internal friction angle and the water content of the mth layer soil layer as C_m、φ_mAnd ω_mThe total cohesive force, the internal friction angle and the water content of the soil body on the excavation surface are respectively C, phi and omega, and the calculation formula is as follows:

C＝∑P_m·C_m φ＝∑P_m·φ_m ω＝∑P_m·ω_m

the geometric characteristics comprise the buried depth of each ring of the shield and the soil mass loss rate under the non-grouting condition, and the buried depth of each ring of the shield is directly obtained through a tunnel design file; the soil mass loss rate under the condition of each ring of non-grouting needs to independently calculate a straight line segment and a curve segment, the straight line segment only considers the volume of a circular ring surrounded by the radius R of a shield cutter head and the outer radius R of a tunnel segment, and the specific calculation formula is as follows:

v＝(R²-r²)πl

in order to deflect the shield when the curve segment is propelled, the over-excavation of the soil body is usually caused, the theoretical soil body loss rate of the shield is greatly different from that of a straight segment, and the specific calculation mode is as follows:

in the formula, R₀The curvature radius of the tunnel, L is the width of a duct piece, L is the length of the shield machine, D is the diameter of the shield machine, delta is the inner side overexcavation amount of the shield machine, delta 'is the theoretical gap in the middle of the shield machine, and delta' is approximately equal to delta;

the process characteristics comprise the type of the shield machine, the length of the shield machine and the diameter of the shield machine, wherein the type of the shield machine is mainly divided into an earth pressure balanced type shield machine and a muddy water balanced type shield machine, and is determined according to the type of the shield machine selected in the actual engineering; the shield length and the shield diameter are obtained according to the shield machine parameter data adopted by specific construction;

the disturbance parameters mainly evaluate indexes of soil disturbance degree of the shield in the tunneling process, and are obtained by analyzing and processing shield real-time tunneling data; dividing the disturbance mechanism into a notch disturbance characteristic, a posture disturbance characteristic and a shield tail disturbance characteristic according to different disturbance mechanisms;

the disturbance condition of the soil body in front of the notch is evaluated by adopting the shield theoretical pressure difference, so that the notch disturbance characteristic is obtained, and the specific calculation mode is as follows:

step S121, calculating according to a geological survey result to obtain the upper load size Q;

s122, calculating a lateral soil pressure coefficient K of a soil body of an excavation surface according to a tunnel geological survey report;

s123, extracting soil pressure data during the propulsion period of each ring, and calculating the average soil pressure P of the ring;

step S124, calculating the theoretical soil pressure difference according to the following formula: Δ P ═ P-Q · K;

the attitude disturbance characteristics comprise the attitude variation of the shield in the horizontal direction and the attitude variation of the shield in the elevation direction; the attitude variation of the shield elevation direction is expressed according to the shield pitch angle variation, namely:

in the formula

For the variation of the shield ring pitch angle,

is the pitch angle at the beginning of each loop,

the pitch angle at the end of each loop;

the shield horizontal direction attitude variation is expressed by adopting horizontal angle variation, and is obtained by calculation according to the deviation of the shield and the axis linearity, and the specific calculation method is as follows:

wherein h'_xIs the amount of horizontal deviation of the notch at the beginning of each loop, h ″_xIs the notch horizontal deviation at the end of each loop, t'_xIs the horizontal deviation of the shield tail at the beginning of each ring, t_xThe shield tail horizontal deviation value at the end of each loop;

the shield tail grouting pressure is obtained by calculating the average pressure value during grouting of each ring of shield tails, and the shield tail grouting filling rate is obtained by calculating the ratio of the ring accumulated grouting amount to the theoretical shield tail clearance, so that the shield tail disturbance characteristic is obtained, wherein the specific calculation formula is as follows:

wherein G is the shield tail grouting filling rate G_tAccumulating grouting amount for the shield tail ring, wherein v is the size of a theoretical shield tail gap;

the settlement type describes the settlement in the shield propulsion process mainly through the settlement form and the settlement amplitude;

the sedimentation form is mainly expressed by the width coefficient of the transverse sedimentation tank, the width coefficient I of the transverse sedimentation tank is obtained by performing curve fitting on transverse sedimentation measuring points in the influence range by using a peck formula and taking the average value, and the peck formula is as follows:

in the formulaS is the sedimentation value of any point on the earth surface; s_maxThe maximum value of the ground surface settlement is positioned right above the axis of the tunnel; x is the horizontal distance between any point and the axis of the tunnel; i is a width coefficient of the transverse settling tank, namely the distance between the axis of the tunnel and the inflection point of the settling tank;

the settlement amplitude comprises a settlement average value of 10m in front of the notch, a settlement average value above the shield body and a settlement average value of 10m behind the shield tail, and the data are obtained by respectively selecting settlement values of axis points of three areas according to the shield stroke and calculating;

the settlement data generator training module is used for extracting the element attributes of the acquired construction data of different projects according to the method, inputting the basic attributes and the disturbance attributes as models, outputting the settlement attributes as models, and training a recurrent neural network model as a settlement data generator; the method comprises the following specific steps:

step S21, data preprocessing, namely converting the metadata into time sequence data, and dividing a training set and a test set;

s22, constructing a three-layer recurrent neural network model;

step S23, training the model by using the training set data, verifying the prediction effect of the model on the test set, and adjusting the model parameters according to the result;

step S24, saving the model;

the settlement data generation module is used for calculating to obtain basic data according to geometric data of practical application projects, geological data and shield tunneling machine parameter data, and then calculating to obtain a disturbance data range by combining a theoretical calculation method;

when the notch disturbance characteristic is calculated, the soil pressure fluctuation in the actual propelling process is simulated according to the theoretical soil pressure value and a certain random fluctuation amount, and the simulated theoretical soil pressure difference is further calculated; when the shield tail disturbance characteristic is calculated, simulating the average grouting pressure according to the grouting pressure range under similar soil layers, and floating the average grouting rate up and down according to the theoretical grouting filling rate of 150%; when the attitude disturbance characteristic is calculated, obtaining theoretical shield elevation and horizontal direction attitude variation according to shield tunnel design axis linearity;

inputting the simulated construction data generated by the method into a trained settlement data generator to obtain the settlement type attribute at each moment; and obtaining the settlement of each point by an empirical settlement curve formula according to the calculated settlement attributes, wherein the concrete steps are as follows:

s31, acquiring geometric data, geological data and shield machine parameter data of the actual construction project tunnel;

s32, designing a simulation construction scheme according to the basic data;

step S33, inputting the simulated construction data into a settlement data generator to obtain settlement data;

step S34, calculating the settlement of each settlement monitoring point at each moment according to the settlement data, wherein the specific calculation method is as follows:

firstly, fitting according to the calculated settlement data and an empirical longitudinal settlement curve formula to obtain a longitudinal settlement curve, wherein the empirical formula is as follows:

in the formula, S_ySettlement of a settlement monitoring point which is at a distance x from the notch on the axis, y is the distance between the monitoring point and the notch, and y is a negative number when the monitoring point is in front of the notch;

and then calculating by combining the width coefficient I of the transverse settling tank to obtain a transverse settling tank formula of the settlement monitoring point:

finally, calculating by combining the two formulas to obtain a settlement value of a settlement monitoring point with the relative incision coordinate of (x, y);

step S35, storing settlement simulation data;

the settlement prediction model pre-training module is used for constructing an LSTM monitoring point settlement prediction model according to shield construction simulation data calculated by a settlement data generator, and the LSTM model is input into the excavation surface soil body cohesive force from t-n to t, the inner friction angle of the excavation surface soil body, the water content of the excavation surface soil body, an upper load, the front soil pressure, the grouting filling rate, the grouting pressure, the shield horizontal direction posture variation, the shield elevation direction posture variation, the longitudinal distance between a settlement monitoring point and a notch, and the transverse distance between the settlement monitoring point and an axis; the output of the model is the accumulated settlement value at the moment t; wherein, the time step n of a single sample of the LSTM model is 12, and the number of LSTM layers is 1;

the real-time settlement prediction module is used for predicting the settlement of the settlement monitoring points in the shield influence range by adopting a settlement prediction model, and specifically comprises the following steps:

s51, acquiring the current shield position, and determining a monitoring point set needing settlement prediction, wherein the monitoring point set specifically comprises transverse and longitudinal monitoring points within a range from ten meters in front of a notch to ten meters behind a shield tail;

s52, acquiring historical shield construction data from t-n to t;

s53, obtaining shield construction parameter setting values of 5 time steps in the future;

step S54, inputting shield construction data from t-n + i to t + i, and predicting the settlement of each monitoring point at the t + i by adopting an LSTM model;

and step S55, repeating the step S54 until the settlement prediction of the future 5 time steps is completed.

2. The shield construction surface subsidence prediction system of claim 1, wherein:

the system further comprises:

the historical construction project data acquisition module is used for collecting shield construction data, tunneling parameter data, geological parameter data and ground settlement data under different construction environments;

the automatic learning module of the settlement prediction model is used for updating the settlement prediction model on line;

the historical construction project data acquisition module is used for acquiring shield construction data of different engineering projects from a shield construction database, classifying the acquired original construction data, extracting data related to settlement prediction in the original construction data, and generating an original data set; the original data set specifically comprises a geological data set, a tunnel geometric data set, a shield tunneling machine parameter data set, a shield tunneling data set and a settlement data set, and each type of data set comprises the following data:

the geological data set comprises: drilling hole positions, starting and stopping burial depth of each soil layer and soil body mechanical parameters of each soil layer; the soil mechanical parameters of each soil layer comprise cohesive force, an internal friction angle, a compression coefficient, a compression modulus, a natural pore ratio, a lateral soil pressure coefficient, a natural water content and a gravity;

the tunnel geometry data set includes: the outer diameter of a tunnel segment, the buried depth of a tunnel and the line type of the tunnel;

the shield machine parameter data set comprises: the diameter of the shield, the length of the shield and the type of the shield process;

the shield tunneling data set includes: shield tunneling stroke, notch horizontal deviation, notch elevation deviation, shield tail horizontal deviation, shield tail elevation deviation, shield slope angle, front soil pressure partition, grouting amount distribution, grouting pressure distribution, tunneling speed, total thrust, slurry type and slurry initial setting time;

the sedimentation data set includes: measuring point position, monitoring time and settlement value;

the automatic learning module of the settlement prediction model is used for carrying out rolling training on the settlement prediction model according to new data generated in a construction site, and the calculation accuracy of the module is improved;

by every interval K_tAnd at each moment, inputting shield parameter data generated newly in comparison with the last training to the module to update the settlement prediction model in the background so as to improve the calculation precision of the model.

3. The method for predicting the shield construction surface subsidence is characterized by comprising the following steps of:

a meta-attribute extraction step, namely recombining the original data sets to extract meta-attributes, and calculating attribute characteristic indexes;

a settlement data generator training step, namely constructing a settlement data generator based on the meta-attributes;

a settlement data generation step, wherein a group of simulation data is generated by combining the engineering characteristics of the current construction project;

a settlement prediction model pre-training step, wherein a settlement prediction model of the current project is trained by combining the generated simulation data to obtain an initial settlement prediction model;

a real-time settlement predicting step, wherein real-time shield tunneling data is obtained to predict settlement values of the surface settlement monitoring points;

in the element attribute extraction step, element attributes are extracted; dividing main factors influencing surface subsidence into three attribute categories of a basic category, a disturbance category and a subsidence category;

each type of element attribute comprises a plurality of tunneling characteristics, the characteristics are obtained by calculating original construction data, and the specific calculation method is as follows:

the foundation type is the tunnel foundation information which can be obtained in the shield tunnel construction stage and comprises soil property, geometric property and process property;

the soil property characteristics specifically comprise cohesive force, an internal friction angle and water content of each ring of tunnel section, and the specific calculation mode of each ring of soil property is as follows:

step S111, calculating the area proportion P of each layer of soil of the ring to the excavation surface respectively_mDefining the burial depth of the boundary on the mth layer of soil layer as d_1，mLower boundary buried depth is d_2，mThe buried depth of the center point of the tunnel section is d, and the radius of the excavation surface is R and S_1，mThe contact area S of the soil body above the initial elevation of the soil layer and the excavation surface_2，mThe contact area between the soil body and the excavation surface is larger than the finished elevation of the soil layer; the specific calculation method comprises the following steps:

step S112, calculating the soil property of the section of the circular tunnel, and defining the cohesive force, the internal friction angle and the water content of the mth layer soil layer as C_m、φ_mAnd w_mThe total cohesive force, the internal friction angle and the water content of the soil body on the excavation surface are respectively C, phi and omega, and the calculation formula is as follows:

C＝∑P_m·C_m φ＝∑P_m·φ_m ω＝∑P_m·ω_m

the geometric characteristics comprise the buried depth of each ring of the shield and the soil mass loss rate under the non-grouting condition, and the buried depth of each ring of the shield can be directly obtained through a tunnel design file; the soil mass loss rate under the condition of each ring of non-grouting needs to independently calculate a straight line segment and a curve segment, the straight line segment only considers the volume of a circular ring surrounded by the radius R of a shield cutter head and the outer radius R of a tunnel segment, and the specific calculation formula is as follows:

v＝(R²-r²)πl

in order to deflect the shield when the curve segment is propelled, the over-excavation of the soil body is usually caused, the theoretical soil body loss rate of the over-excavation is greatly different from that of a straight segment, and the specific calculation mode comprises the following steps:

in the formula, R₀The curvature radius of the tunnel, L is the width of a duct piece, L is the length of the shield machine, D is the diameter of the shield machine, delta is the inner side overexcavation amount of the shield machine, delta 'is the theoretical gap in the middle of the shield machine, and delta' is approximately equal to delta;

the process characteristics comprise the type of the shield machine, the length of the shield machine and the diameter of the shield machine, wherein the type of the shield machine is mainly divided into an earth pressure balanced type shield machine and a muddy water balanced type shield machine, and is determined according to the type of the shield machine selected in the actual engineering; the shield length and the shield diameter are obtained according to the shield machine parameter data adopted by specific construction;

the disturbance parameters mainly evaluate indexes of soil disturbance degree of the shield in the tunneling process, and are obtained by analyzing and processing shield real-time tunneling data; dividing the disturbance mechanism into a notch disturbance characteristic, a posture disturbance characteristic and a shield tail disturbance characteristic according to different disturbance mechanisms;

the disturbance condition of the soil body in front of the notch is evaluated by adopting the shield theoretical pressure difference, so that the notch disturbance characteristic is obtained, and the specific calculation mode comprises the following steps:

step S121, calculating according to a geological survey report to obtain the upper load size Q;

s122, calculating a lateral soil pressure coefficient K of a soil body of an excavation surface according to a tunnel geological survey report;

s123, extracting soil pressure data during the propelling of the ring, and calculating the average soil pressure P of the ring;

step S124, calculating the theoretical soil pressure difference according to the following formula: Δ P ═ P-Q · K;

the attitude disturbance characteristics comprise the attitude variation of the shield in the horizontal direction and the attitude variation of the shield in the elevation direction; the attitude variation of the shield elevation direction is expressed according to the shield pitch angle variation, namely:

in the formula (I), the compound is shown in the specification,

for the variation of the shield ring pitch angle,

is the pitch angle at the beginning of each loop,

the pitch angle at the end of each loop;

the shield horizontal direction attitude variation is expressed by adopting horizontal angle variation, and is obtained by calculation according to the deviation of the shield and the axis linearity, and the specific calculation method is as follows:

wherein h'_xIs the amount of horizontal deviation of the notch at the beginning of each loop, h ″_xIs the notch horizontal deviation at the end of each loop, t'_xIs the horizontal deviation of the shield tail at the beginning of each ring, t_xThe shield tail horizontal deviation value at the end of each loop;

the shield tail grouting pressure is obtained by calculating the average pressure value during grouting of each ring of shield tails, and the shield tail grouting filling rate is obtained by calculating the ratio of the ring accumulated grouting amount to the theoretical shield tail clearance, so that the shield tail disturbance characteristic is obtained, wherein the specific calculation formula is as follows:

wherein G is the shield tail grouting filling rate G_tAccumulating grouting amount for the shield tail ring, wherein v is the size of a theoretical shield tail gap;

the settlement type describes the settlement in the shield propulsion process mainly through the settlement form and the settlement amplitude;

the sedimentation form is mainly expressed by the width coefficient of the transverse sedimentation tank, the width coefficient I of the transverse sedimentation tank is obtained by performing curve fitting on transverse sedimentation measuring points in the influence range by using a peck formula and taking the average value, and the peck formula is as follows:

in the formula, S is the sedimentation value of any point on the earth surface; s_maxThe maximum value of the ground surface settlement is positioned right above the axis of the tunnel; x is the horizontal distance between any point and the axis of the tunnel; i is a width coefficient of the transverse settling tank, namely the distance between the axis of the tunnel and the inflection point of the settling tank;

the settlement amplitude comprises a settlement average value of 10m in front of the notch, a settlement average value above the shield body and a settlement average value of 10m behind the shield tail, and the data are obtained by respectively selecting settlement values of axis points of three areas according to the shield stroke and calculating;

in the step of training the settlement data generator, after extracting the element attributes of the acquired construction data of different projects according to the method, taking the basic class attributes and the disturbance class attributes as model inputs, taking the settlement class attributes as model outputs, and training a recurrent neural network model as the settlement data generator; the method comprises the following specific steps:

step S21, data preprocessing, namely converting the metadata into time sequence data, and dividing a training set and a test set;

s22, constructing a three-layer recurrent neural network model;

step S23, training the model by using the training set data, verifying the prediction effect of the model on the test set, and adjusting the model parameters according to the result;

step S24, saving the model;

in the sedimentation data generation step, basic data are obtained through calculation according to geometric data of practical application projects, geological data and shield tunneling machine parameter data, and then a disturbance data range is obtained through calculation by combining a theoretical calculation method;

when the notch disturbance characteristic is calculated, the soil pressure fluctuation in the actual propelling process is simulated according to the theoretical soil pressure value and a certain random fluctuation amount, and the simulated theoretical soil pressure difference is further calculated; when the shield tail disturbance characteristic is calculated, simulating the average grouting pressure according to the grouting pressure range under similar soil layers, and floating the average grouting rate up and down according to the theoretical grouting filling rate of 150%; when the attitude disturbance characteristic is calculated, obtaining theoretical shield elevation and horizontal direction attitude variation according to shield tunnel design axis linearity;

inputting the simulated construction data generated by the method into a trained settlement data generator to obtain the settlement type attribute at each moment; and obtaining the settlement of each point by an empirical settlement curve formula according to the calculated settlement attributes, wherein the concrete steps are as follows:

s31, acquiring geometric data, geological data and shield machine parameter data of the actual construction project tunnel;

s32, designing a simulation construction scheme according to the basic data;

step S33, inputting the simulated construction data into a settlement data generator to obtain settlement data;

step S34, calculating the settlement of each settlement monitoring point at each moment according to the settlement data, wherein the specific calculation method is as follows:

firstly, fitting according to the calculated settlement data and an empirical longitudinal settlement curve formula to obtain a longitudinal settlement curve, wherein the empirical formula is as follows:

in the formula, S_ySettlement of a settlement monitoring point which is at a distance x from the notch on the axis, y is the distance between the monitoring point and the notch, and y is a negative number when the monitoring point is in front of the notch;

and then calculating by combining the width coefficient I of the transverse settling tank to obtain a transverse settling tank formula of the settlement monitoring point:

finally, calculating by combining the two formulas to obtain a settlement value of a settlement monitoring point with the relative incision coordinate of (x, y);

step S35, storing settlement simulation data;

in the settlement prediction model pre-training step, constructing an LSTM monitoring point settlement prediction model according to shield construction simulation data calculated by a settlement data generator, wherein the LSTM model is input into the excavation surface soil body cohesive force from t-n to t, the inner friction angle of the excavation surface soil body, the water content of the excavation surface soil body, an upper load, the front soil pressure, the grouting filling rate, the grouting pressure, the shield horizontal direction posture variation, the shield elevation direction posture variation, the longitudinal distance between a settlement monitoring point and a notch, and the transverse distance between the settlement monitoring point and an axis; the output of the model is the accumulated settlement value at the moment t; wherein, the time step n of a single sample of the LSTM model is 12, and the number of LSTM layers is 1;

in the real-time settlement predicting step, a settlement predicting model is adopted to predict the settlement of the settlement monitoring points in the shield influence range, and the method specifically comprises the following steps:

s51, acquiring the current shield position, and determining a monitoring point set needing settlement prediction, wherein the monitoring point set specifically comprises transverse and longitudinal monitoring points within a range from ten meters in front of a notch to ten meters behind a shield tail;

s52, acquiring historical shield construction data from t-n to t;

s53, obtaining shield construction parameter setting values of 5 time steps in the future;

step S54, inputting shield construction data from t-n + i to t + i, and predicting the settlement of each monitoring point at the t + i by adopting an LSTM model;

and step S55, repeating the step S54 until the settlement prediction of the future 5 time steps is completed.

4. The shield construction ground surface subsidence prediction method of claim 3, wherein:

the method further comprises the following steps:

acquiring historical construction project data, namely collecting shield construction data, tunneling parameter data, geological parameter data and ground settlement data under different construction environments;

a settlement prediction model automatic learning step, which is used for updating the settlement prediction model on line;

in the historical construction project data acquisition step, shield construction data of different engineering projects are acquired from a shield construction database, the acquired original construction data are classified, data related to settlement prediction are extracted, and an original data set is generated; the original data set specifically comprises a geological data set, a tunnel geometric data set, a shield tunneling machine parameter data set, a shield tunneling data set and a settlement data set, and each type of data set comprises the following data:

the geological data set comprises: drilling hole positions, starting and stopping burial depth of each soil layer and soil body mechanical parameters of each soil layer; the soil mechanical parameters of each soil layer comprise cohesive force, an internal friction angle, a compression coefficient, a compression modulus, a natural pore ratio, a lateral soil pressure coefficient, a natural water content and a gravity;

the tunnel geometry data set includes: the outer diameter of a tunnel segment, the buried depth of a tunnel and the line type of the tunnel;

the shield machine parameter data set comprises: the diameter of the shield, the length of the shield and the type of the shield process;

the shield tunneling data set includes: shield tunneling stroke, notch horizontal deviation, notch elevation deviation, shield tail horizontal deviation, shield tail elevation deviation, shield slope angle, front soil pressure partition, grouting amount distribution, grouting pressure distribution, tunneling speed, total thrust, slurry type and slurry initial setting time;

the sedimentation data set includes: measuring point position, monitoring time and settlement value;

in the automatic learning step of the settlement prediction model, rolling training is carried out on the settlement prediction model according to new data generated in a construction site, and the calculation accuracy of the module is improved;

by every interval K_tAnd at each moment, inputting shield parameter data generated newly in comparison with the last training to the module to update the settlement prediction model in the background so as to improve the calculation precision of the model.

Images