CN102998367B - Damage identification method based on virtual derivative structure - Google Patents

Abstract

The invention discloses a damage identification method based on a virtual derived structure, which includes the following steps: first, a finite element model of the structure to be tested (also called a source structure) in an undamaged state is established by finite element software; The test system measures the first few modal data of the vibration of the source structure; then, constructs a series of virtual derived structures based on the source structure and calculates the corresponding characteristic frequency parameters; finally, synthesizes the characteristics of the source structure and all virtual derived structures Frequency, through the hybrid frequency sensitivity diagnostic program, the source structure damage identification results including damage location and degree are obtained. The technical solution has simple test equipment and low cost, effectively overcomes the shortcomings of existing test technologies, greatly improves the accuracy of damage identification, and is especially suitable for damage identification of large structures.

Description

A Damage Recognition Method Based on Virtual Derived Structure

technical field

The invention belongs to the structural damage recognition field of the civil engineering discipline, and relates to a damage recognition method based on a virtual derived structure.

Background technique

With the rapid development of the national economy, a large number of large-scale civil engineering structures have been built, such as super high-rise buildings, long-span bridges, and long-span space roof trusses. Due to the increase in the service life and the influence of environmental corrosion and disaster loads, these structures will inevitably be damaged. Partial damage to the structure may lead to rapid damage to the entire structure, causing huge losses to people's lives and property. To make timely and accurate identification of the damage status of the system, so as to facilitate maintenance and reinforcement to avoid catastrophic consequences.

Traditional damage identification methods are mostly visible local experimental methods, such as ultrasonic methods, magnetic field methods, and temperature field methods. However, the above-mentioned methods all have the following main defects: firstly, the above-mentioned methods require the approximate location of the damage to be known in advance, so the predictability is poor; The above method has the above defects, so it is not suitable for damage identification of large civil engineering structures.

In recent years, the study of holistic, real-time, and applicable methods for large-scale structural damage identification has become a common focus of attention in many engineering and technical fields such as civil engineering, machinery, and aviation. Among them, the method of using the change of vibration modal data of large structures to inversely identify its damage is a new technology at present. The basic principle of this type of method is: structural vibration modal data is a function of structural physical parameters (such as mass, stiffness, etc.), so changes in structural physical parameters (structural damage) will inevitably cause structural vibration modal parameters (frequency and mode shape) The changes of these vibration data can be used to identify the damage condition of the structure in turn. However, due to the limitations of testing techniques, only the first few low-order modes of structural vibration can be measured at present, which is far from meeting the needs of large-scale structural damage identification. The number of identification equations that can be established is far less than the number of unknown damage parameters, which is one of the most critical reasons for the failure of current methods for damage identification.

In view of this, the inventor combined his years of experience in the field of structural damage identification to conduct long-term research on the defects in the above-mentioned technical field, and this case came about.

Contents of the invention

The main purpose of the present invention is to provide a damage identification method based on a virtual derived structure, which is easy to operate, simple in testing equipment, low in cost, and improves the accuracy of damage identification. the

In order to achieve the above object, the technical scheme of the present invention is as follows:

A damage identification method based on a virtual derived structure, including the following steps: firstly, the finite element model of the undamaged state of the structure to be tested (also called the source structure) is established by finite element software; secondly, the finite element model is obtained by measuring the For the data of the first few modals (i.e., characteristic frequencies and corresponding mode shapes) of the vibration of the source structure, the modal order to be measured is determined based on comprehensive consideration of the complexity of the structure, the accuracy of the measuring instrument, and the conditions of the test site; then, with Based on the source structure, a series of virtual derived structures are constructed and the corresponding characteristic frequency parameters are calculated; finally, the characteristic frequencies of the source structure and all virtual derived structures are integrated, and the source structure including the damage location and degree is obtained through the mixed frequency sensitivity diagnostic program. Structural damage identification results.

The modal test system includes a wireless data acquisition instrument, a plurality of wireless sensors and a set of modal analysis software, and the first few order characteristic frequencies and corresponding mode shape data of the vibration of the source structure are obtained by the modal test system; wherein The wireless sensors are arranged in the parts of the structure to be tested that are prone to damage, and the number of wireless sensors according to the formula to determine, where is the total number of elements in the finite element model, is the number of measured characteristic frequencies.

Further, the modal data of the first few orders refers to the modal data between the first 4 to 10 orders.

Further, the virtual derived structure can be obtained by virtual arrangement of mass or stiffness on the source structure. The virtual mass or stiffness is placed at the location of the wireless sensor, which can be arranged alone or in combination. The size of the virtual mass or stiffness It is taken as 10%~15% of the value of the mass matrix or stiffness matrix in the finite element model corresponding to the position of the sensor in the undamaged state of the source structure; a series of virtual derived structures can be obtained by using different virtual layout schemes, and the constructed Total number of virtual derived structures and number of sensors The relationship between ; After the virtual derived structure is constructed, the first-order frequency sensitivity formula is used to calculate the first few order characteristic frequency parameters of the virtual derived structure.

Further, the mixed sensitivity diagnostic program includes the following steps: firstly, according to the finite element model of the source structure and the virtual derived structure in the undamaged state, respectively calculate the front of the source structure and the virtual derived structure in the undamaged state order eigenfrequency and the corresponding first-order frequency sensitivity; then the measured source structure is The order eigenfrequency and the front of each virtual derived structure The first-order sensitivity equations of frequency changes before and after damage are listed together, and all unknown damage parameters are solved by generalized inverse technology ; Finally, the damage location and degree of the source structure can be accurately identified by outputting the calculated damage parameters in the form of a histogram or the like.

In this technical solution, only a few sensors are used to measure the first few modes of vibration of the structure to be measured (source structure), and more frequency information is obtained by constructing a series of virtual derived structures, and finally the source structure and all derived structures are combined frequency to identify the damage condition of the source structure.

By adopting this technical solution, the following beneficial effects are obtained: first, the present invention only needs a small number of sensors, and only the first few low-order modes of structural vibration can be measured, and the test equipment is simple and low in cost; second, The present invention obtains more frequency parameters for structural damage identification by virtualizing a series of derived structures, effectively overcomes the deficiencies of existing testing techniques, thereby greatly improving the accuracy of damage identification, and is especially suitable for large-scale structures In damage identification; third, the derived structure of the present invention is constructed virtually, so there is no need to actually arrange mass or stiffness components on the structure to be tested, only conventional modal testing equipment and a small number of sensors are required, so the operation is convenient It is easy to implement without adding additional testing workload; fourth, the number of virtual derivative structures that can be constructed by the present invention grows rapidly according to the speed of the power as the number of sensors increases, so the present invention meets the needs of damage identification The invention uses a minimum number of sensors.

In order to further explain the technical solution of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments.

Description of drawings

Fig. 1 is the implementation schematic diagram of the present invention; Wherein label explanation: 1, wireless acceleration sensor; 2, wireless data acquisition instrument; 3, modal analysis software; 1, 2 and 3 form modal test system jointly; 4, to-be-tested structure (source structure); 5 is the finite element model of the source structure in the undamaged state; 6 is the virtual derived structure; 7 is the mixed sensitivity diagnostic program; 8 is the damage identification result.

Fig. 2 is a grid structure model to be detected;

Figure 3 is the identification result of a single damage situation (unit 15 loses 15% of its stiffness);

Figure 4. Identification results of multiple damage situations (units 7 and 15 both lose 15% of stiffness).

Detailed ways

The implementation of the present invention will be further described in detail below in conjunction with accompanying drawings 1 to 4 .

A damage identification method based on a virtual derived structure, including the following steps: firstly, the finite element model of the undamaged state of the structure to be tested (also called the source structure) is established by finite element software; secondly, the finite element model is obtained by measuring the For the data of the first few modals (i.e., characteristic frequencies and corresponding mode shapes) of the vibration of the source structure, the modal order to be measured is determined based on the complexity of the structure, the accuracy of the measuring instrument and the conditions of the test site. In this implementation In the example, it is taken as the first 4~10 orders; then, a series of virtual derived structures are constructed based on the source structure and the corresponding eigenfrequency parameters are calculated; finally, the eigenfrequencies of the source structure and all virtual derived structures are integrated, through A hybrid frequency sensitivity diagnostic procedure yields damage identification results for source structures including damage location and extent.

The finite element model of the source structure in an undamaged state can be obtained by modeling the source structure with general finite element software (such as general commercial software such as ANSYS or MATLAB); details will not be repeated here. In this embodiment, the finite element model includes a stiffness matrix and a mass matrix, does not consider structural damping, and uses a lumped mass matrix. the

The modal test system includes a wireless data acquisition instrument, multiple wireless acceleration sensors and a set of modal analysis software (all of which can be purchased from the market), and the first few orders of vibration of the source structure are measured by the modal test system Eigenfrequency and corresponding mode shape data. Among them, the wireless sensors are arranged in the parts that are prone to damage in the structure to be tested (determined according to engineering experience, for example, the beam structure can be arranged at the mid-span and beam end section, and the plate structure can be arranged at the four corners and the center position), the number of wireless sensors according to the formula to determine, where is the total number of elements in the finite element model, is the number of measured characteristic frequencies.

The virtual derived structure can be obtained by virtually arranging mass or stiffness on the source structure (not actually arranging mass or stiffness on the source structure). The virtual mass or stiffness is placed at the location of the wireless sensor, which can be arranged separately It can also be arranged in combination. The size of the virtual mass or stiffness is generally taken as 10%~15% of the value corresponding to the position of the sensor in the mass matrix or stiffness matrix in the finite element model of the source structure in an undamaged state. A series of virtual derived structures can be obtained by using different virtual layout schemes, and the total number of virtual derived structures that can be constructed and the number of sensors The relationship between . After the virtual derivative structure is constructed, the first-order frequency sensitivity formula is used to calculate the first few order characteristic frequency parameters of the virtual derivative structure.

Next, take the method of virtual arrangement of quality as an example to construct a virtual derived structure and calculate the corresponding eigenfrequency. Let the mass matrix and stiffness matrix divisions of the source structure in the undamaged state be and , they can be obtained by modeling the structure with finite element software, where the mass matrix is a diagonal matrix

(1)

In formula (1) Indicates the total number of degrees of freedom. Before and after the source structure is damaged, the mass generally remains the same but only the stiffness is lost. Therefore, the finite element model of the source structure in the damaged state is: the mass matrix remains , while the stiffness matrix is reduced to , is the unknown stiffness matrix, which can be expressed as

(2)

in and respectively in the finite element model The damage parameters and element stiffness matrix of each element. Without loss of generality, assuming that a total of 2 acceleration sensors are arranged on the source structure, which are respectively arranged at the first and second degrees of freedom, then three virtual derived structures can be constructed by virtual arrangement of masses at these two sensor arrangement points , respectively: the first virtual derived structure is a virtual arrangement of a lumped mass at the first degree of freedom alone , the stiffness matrix of the resulting virtual derived structure is still the same as that of the source structure, and its mass matrix is the source structure mass matrix with the additional virtual mass matrix the sum of

, (3), (4)

The second virtual derived structure is a virtual arrangement of a concentrated mass at the second degree of freedom alone , the stiffness matrix of the resulting virtual derived structure is still the same as that of the source structure, and its mass matrix is the source structure mass matrix with the additional virtual mass matrix the sum of

, (5), (6)

The third virtual derived structure is the virtual arrangement of 2 concentrated masses at the 1st and 2nd degrees of freedom at the same time and , the stiffness matrix of the resulting virtual derived structure is still the same as that of the source structure, and its mass matrix is the source structure mass matrix with the additional virtual mass matrix the sum of

, (7),(8)

After the above three virtual derived structures are constructed, the corresponding eigenfrequency values can be calculated by the first-order frequency sensitivity formula respectively, and the calculation formula is as follows:

,in , (9)

In formula (9) and The first measurements obtained for the source structure with the modal test system order eigenfrequencies and mode shapes, for the first The first virtual derived structure order characteristic frequency. noticed In addition to the position where the virtual mass is arranged, the other elements are all 0, so the calculation of formula (9) does not require the measurement mode shape with complete degrees of freedom , as long as there are partial mode shape values corresponding to the position of the virtual mass, which happens to be the placement position of the sensor, so all the data used in the calculation with formula (9) can be obtained by measurement.

The mixed sensitivity diagnostic program includes the following steps: firstly, according to the finite element model of the source structure and the virtual derived structure in the undamaged state, respectively calculate the front order eigenfrequency and the corresponding first-order frequency sensitivity; then the measured source structure is The order eigenfrequency and the front of each virtual derived structure The first-order sensitivity equations of frequency changes before and after damage are listed together, and all unknown damage parameters are solved by generalized inverse technology ; Finally, the damage location and degree of the source structure can be accurately identified by outputting the calculated damage parameters in the form of a histogram or the like.

The formula used in the mixed sensitivity diagnostic program is as follows: by the model matrix in the undamaged state of the source structure and , the eigenfrequency of the source structure in the undamaged state can be obtained by solving the following generalized eigenvalue equation

(10)

in and Respectively, when the source structure is undamaged characteristic frequency and mode shape. No. The eigenfrequency about the The formula for calculating the first-order sensitivity of a damage parameter is

(11)

Similarly, No. The eigenfrequency and the corresponding first-order sensitivity calculation formula of a virtual derived structure in the undamaged state are as follows:

, (12),(13)

According to equations (10)-(13), the front of the source structure and all virtual derived structures in the undamaged state can be calculated. The first-order eigenfrequency and the corresponding first-order frequency sensitivity. The measured source structure is then placed before order eigenfrequency and the front The first-order sensitivity equations of all frequency changes before and after damage are listed together, which is

(14)

where the frequency change vector for

(15)

damage parameter vector for

(16)

Mixed Sensitivity Matrix for

(17)

All damage parameters can be solved by generalized inverse technique from equation (14), namely

(18)

The superscript "+" in Equation (18) represents the generalized inverse of the matrix. Finally, all damage parameters calculated by equation (18) are output in the form of a histogram, which can accurately identify whether the source structure is damaged, the location and degree of damage.

Figure 2 is a grid structure model to be tested. The basic parameters of the structure are: elastic modulus E=200GP _a , density ρ=7.8×10 ³ Kg/m ³ , length of each member L=1m, transverse Cross-sectional area A=0.004m ² . Two damage situations are simulated. The first case is a single damage: assuming that the stiffness of member 15 in Fig. 2 is lost by 15%; the second case is multiple damages: assuming that both members 7 and 15 in Fig. 2 lose 15% of their stiffness.

The steps for damage identification of the grid structure shown in Figure 2 using the method proposed in this case are as follows: First, analyze the structural model to know the total number of elements in the finite element model , if only the first four modes are measured (ie ), then according to the formula The number of wireless sensors can be determined , so two horizontal and vertical wireless acceleration sensors are arranged on node A in Figure 2, and one horizontal acceleration sensor is arranged on node B, a total of three sensors are arranged; secondly, the modal test system (1+2+3) is used to measure The first four order eigenfrequencies of grid structure vibration and the corresponding mode shapes; third, use finite element software to establish the finite element model 5 of the undamaged state of the source structure; fourth, the horizontal and vertical The two directions (that is, the arrangement directions of the two sensors of node A) are virtually arranged respectively, and the size of the virtual mass is taken as 10% of the mass value of the finite element model corresponding to the position of the sensor, so as to obtain two virtual derived structures 6, And the first four order eigenfrequencies of each virtual derived structure are calculated by formula (9); fifth, the measured source structure front The order eigenfrequency and the front of each virtual derived structure The order eigenfrequencies are integrated together and input into the mixed sensitivity diagnostic program 7 to output the damage identification result 8 of the source structure.

The damage identification results using the method proposed in this case are listed in Fig. 3 and Fig. 4 under two kinds of damage situations. It can be seen from Fig. 3 that for a single damage case, the damage identification result of the proposed method in this case clearly shows that the rod 15 is damaged, and the damage parameter value is 13.9%, which is very close to the assumed value of 15%. It can be seen from Figure 4 that for multiple damage situations, the damage identification results of the proposed method in this case also clearly show that members 7 and 15 are damaged, and the damage parameter values are 12.4% and 13.7%, respectively, which are very close to the assumed value of 15%. near. In summary, the damage identification method proposed in this case is easy to operate and implement, and the identification results are accurate and reliable.

This technical solution only uses a few sensors to measure the first few modes of the vibration of the structure to be tested (source structure), and obtains more frequency information by constructing a series of virtual derived structures, and finally combines the frequencies of the source structure and all derived structures to obtain The damage condition of the source structure is identified. In the foregoing technical solutions, the technical contents not specifically described in the present invention are all prior art, and will not be repeated here.

The invention only needs a small number of sensors, and only the first few low-order modes of structural vibration can be measured. The test equipment is simple and low in cost; more frequency parameters are obtained by virtualizing a series of derived structures for Structural damage identification effectively overcomes the deficiencies of existing testing technologies, greatly improves the accuracy of damage identification, and is especially suitable for damage identification of large structures; the derived structure of the present invention is constructed virtually, so there is no need to The actual arrangement of mass or stiffness components in the structure can be carried out only by conventional modal test equipment and a small number of sensors, so the operation is convenient and easy to implement, and there is no additional test workload; the number of virtual derivative structures that can be constructed by the present invention As the number of sensors increases, the number of sensors increases rapidly according to the power of the power, so the number of sensors used in the present invention is the least under the premise of meeting the damage identification requirements.

The above descriptions are only specific embodiments of the present invention, and are not limitations to the design of this case. All equivalent changes made according to the design key of this case all fall within the scope of protection of this case.

Claims (1)

1. based on a damnification recognition method for virtual derivative structure, it is characterized in that: comprise the following steps: first, set up the finite element model treated under the non-faulted condition of geodesic structure by finite element software; Secondly, obtained former rank modal data of source structure vibration by mould measurement systematic survey, need the rank number of mode measured to consider according to the complexity of structure, the precision of surveying instrument and test site condition and determines; Then, based on source structure, construct a series of virtual derivative structure and calculate corresponding characteristic frequency parameter; Finally, the characteristic frequency of comprehensive source structure and all virtual derivative structure, by hybrid frequency sensitivity diagnostic routine, draws the source structure non-destructive tests result comprising damage position and degree;

Described mould measurement system comprises a data acquisition instrument, multiple wireless senser and a set of model analysis software, obtains former rank characteristic frequency of source structure vibration and corresponding Data of Mode by mould measurement systematic survey; Wherein wireless senser is arranged in the position treating damage easily occurs in geodesic structure, and the quantity of wireless senser can by formula determine, wherein N is the sum of unit in finite element model, and y is the number of the characteristic frequency measured;

Described former rank modal data, refers to modal data between front 4-10 rank;

Described virtual derivative structure can be obtained by the way of virtual arrangement quality or rigidity on source structure, the position of virtual mass or rigidity is the position at wireless senser place, can arrange separately and also can combine layout, the size of virtual mass or rigidity to be generally taken as under the non-faulted condition of source structure mass matrix or stiffness matrix in finite element model and to correspond to 10% ~ 15% of sensing station place numerical value; Adopt different virtual arrangement schemes just can obtain a series of virtual derivative structure, the pass between the total z of the virtual derivative structure that can construct and the quantity x of sensor is ; After virtual derivative structure constructs, then adopt fundamental frequency sensitivity formula to calculate the former rank characteristic frequency parameter obtaining virtual derivative structure;

Described Mixed Sensitivity diagnostic routine, comprise the following steps: first, according to the finite element model under source structure and the non-faulted condition of virtual derivative structure, calculate the front y rank characteristic frequency of source structure and virtual derivative structure under non-faulted condition and corresponding fundamental frequency sensitivity respectively; Then y rank characteristic frequency before y rank characteristic frequency before the source structure of measurement gained and each virtual derivative structure is combined, list the one order equation of frequency variation before and after damage, solve all unknown impairment parameter by generalizde inverse ; Finally by forms such as histograms, the impairment parameter calculating gained is exported damage position and the degree that accurately can identify source structure.