[go: up one dir, main page]

CN109255490B - A prediction method of external corrosion rate of buried pipeline based on KPCA-BAS-GRNN - Google Patents

A prediction method of external corrosion rate of buried pipeline based on KPCA-BAS-GRNN Download PDF

Info

Publication number
CN109255490B
CN109255490B CN201811143056.7A CN201811143056A CN109255490B CN 109255490 B CN109255490 B CN 109255490B CN 201811143056 A CN201811143056 A CN 201811143056A CN 109255490 B CN109255490 B CN 109255490B
Authority
CN
China
Prior art keywords
corrosion
prediction model
neural network
regression neural
generalized regression
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201811143056.7A
Other languages
Chinese (zh)
Other versions
CN109255490A (en
Inventor
骆正山
姚梦月
骆济豪
王小完
田珮琦
秦越
黄仁惠
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian University of Architecture and Technology
Original Assignee
Xian University of Architecture and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian University of Architecture and Technology filed Critical Xian University of Architecture and Technology
Priority to CN201811143056.7A priority Critical patent/CN109255490B/en
Publication of CN109255490A publication Critical patent/CN109255490A/en
Application granted granted Critical
Publication of CN109255490B publication Critical patent/CN109255490B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
    • G01N27/26—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating electrochemical variables; by using electrolysis or electrophoresis
    • G01N27/416—Systems
    • G01N27/4166—Systems measuring a particular property of an electrolyte
    • G01N27/4168—Oxidation-reduction potential, e.g. for chlorination of water
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/14—Pipes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/04—Ageing analysis or optimisation against ageing

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Marketing (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Chemical & Material Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Molecular Biology (AREA)
  • Pathology (AREA)
  • Geometry (AREA)
  • Evolutionary Computation (AREA)
  • Immunology (AREA)
  • Computer Hardware Design (AREA)
  • General Health & Medical Sciences (AREA)
  • Biochemistry (AREA)
  • Analytical Chemistry (AREA)
  • Electrochemistry (AREA)
  • Chemical Kinetics & Catalysis (AREA)
  • Testing Resistance To Weather, Investigating Materials By Mechanical Methods (AREA)

Abstract

本发明公开了一种基于KPCA‑BAS‑GRNN的埋地管道外腐蚀速率预测方法，包括以下步骤：将管道外腐蚀指标体系的数据分为训练样本及测试样本，再通过训练样本对广义回归神经网络的腐蚀预测模型进行训练，并采用天牛须搜索算法优化腐蚀预测模型的光滑因子σ，然后将测试样本输入到训练后的广义回归神经网络的腐蚀预测模型中，得预测值，再根据预测值及实测值判断训练后广义回归神经网络的腐蚀预测模型的优劣性，当训练后广义回归神经网络的腐蚀预测模型较优时，则将待预测埋地管道外腐蚀的检测数据输入到训练后的广义回归神经网络的腐蚀预测模型中，得埋地管道的外腐蚀速率，该方法能够实现管道腐蚀速率的预测，并且可靠性较高。

Figure 201811143056

The invention discloses a method for predicting the external corrosion rate of buried pipelines based on KPCA-BAS-GRNN. The method includes the following steps: dividing the data of the external corrosion index system of the pipelines into training samples and test samples, and then using the training samples to perform a generalized regression neural The corrosion prediction model of the network is trained, and the smooth factor σ of the corrosion prediction model is optimized by the beetle search algorithm, and then the test samples are input into the corrosion prediction model of the trained generalized regression neural network to obtain the predicted value, and then according to the prediction value and measured value to judge the pros and cons of the corrosion prediction model of the generalized regression neural network after training. When the corrosion prediction model of the generalized regression neural network after training is better, the detection data of the external corrosion of the buried pipeline to be predicted is input into the training In the corrosion prediction model of the latter generalized regression neural network, the external corrosion rate of the buried pipeline is obtained. This method can realize the prediction of the corrosion rate of the pipeline and has high reliability.

Figure 201811143056

Description

KPCA-BAS-GRNN-based buried pipeline external corrosion rate prediction method
Technical Field
The invention belongs to the field of corrosion protection technology of oil and gas pipelines, and relates to a KPCA-BAS-GRNN-based buried pipeline external corrosion rate prediction method.
Background
After the reform is opened, Chinese oil and gas pipeline enterprises also enter the stage of the reform development, a large number of foreign oil and gas pipeline related advanced technologies are introduced, and the foreign advanced management experience is actively learned, so that the overall level of the construction of the oil and gas pipelines in China is obviously improved. The Chuanqi east transportation engineering is another natural gas long-distance pipeline network transportation engineering following the West-east transportation engineering of China. The total investment is 626.76 hundred million yuan, the general plain gas field of Dazhou, Sichuan, Chongqing, Hubei, Jiangxi, Anhui, Jiangsu, Zhejiang and Shanghai 6 province in West province spans 2 cities, the total length of the pipeline is 2170 kilometer, and the natural gas is transported 120 billions of cubic meters in a year, which is equivalent to 1/7 of the natural gas consumption in China in 2009. To date, China has formed a basic pipeline network traversing east-west and longitudinal north and south, and has very important significance for ensuring stable transportation of oil and gas fields, meeting the requirements of industrial production and people life on oil and gas energy, and ensuring sustainable, stable and balanced development of social economy.
In the process of conveying the buried pipeline, the buried pipeline is influenced by a plurality of corrosion factors due to complex geological conditions, and can be divided into four corrosion factors of physics, chemistry, electrochemistry and microorganisms according to properties. Among the physical factors are: water content, volume weight, porosity, temperature, texture, etc.; the chemical factors include: pH, sulfate content, chloride content, sulfide content, and the like; the electrochemical corrosion mainly comprises: soil resistivity, potential gradient, redox potential, etc.; the microbial corrosion includes: sulfate reducing bacteria, neutral sulfurous bacteria, etc. With the increase of the service life, the corrosion depth of the pipeline can be gradually increased under the corrosion factors, so that the corrosion rate of the pipeline can be accurately predicted, effective protective measures can be taken in time, the pipeline safety accidents can be reduced, and a theoretical basis is provided for the updating and maintenance work of the pipeline.
Statistically, pipeline accidents 10798 occur in 1992 and 2012 in the United states, wherein the accidents caused by corrosion account for 18.5%; corrosion leakage accidents account for more than half of all pipeline accidents occurring in 2000 + 2012 of canada. The corrosion problem of oil and gas pipelines in China is also prominent, and domestic scholars find that 39.5 percent of accidents are caused by pipeline corrosion through statistics of natural gas pipeline accidents in Sichuan province; in a particularly serious accident of leakage and explosion of Donghuang oil pipelines in Qingdao in 2013, investigation shows that corrosion is one of important reasons for perforation and leakage of the pipelines. Therefore, establishing a reliable pipeline corrosion rate prediction model has important significance for reducing accidents and avoiding casualties.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a KPCA-BAS-GRNN-based buried pipeline external corrosion rate prediction method, which can realize the prediction of pipeline corrosion rate and has higher reliability.
In order to achieve the purpose, the method for predicting the corrosion rate outside the buried pipeline based on KPCA-BAS-GRNN comprises the following steps:
1) acquiring detection data of corrosion outside the buried pipeline, and constructing an external pipeline corrosion index system according to the detection data of the corrosion outside the buried pipeline;
2) constructing a corrosion prediction model of the generalized regression neural network;
3) dividing the data of the pipeline external corrosion index system obtained in the step 1) into a training sample and a test sample, training a corrosion prediction model of the generalized regression neural network through the training sample, optimizing a smooth factor sigma of the corrosion prediction model by adopting a Tianniu Lexus search algorithm, inputting the test sample into the trained corrosion prediction model of the generalized regression neural network to obtain a predicted value, judging the superiority and inferiority of the trained corrosion prediction model of the generalized regression neural network according to the predicted value and an actual measurement value, inputting the detection data of the external corrosion of the buried pipeline to be predicted into the trained corrosion prediction model of the generalized regression neural network to obtain the external corrosion rate of the buried pipeline when the trained corrosion prediction model of the generalized regression neural network is superior, and when the trained corrosion prediction model of the generalized regression neural network is inferior, go to step 1).
Let training sample X ═ X1，x2，…，xn}，xi∈Rp，RpFor the input space, P is the dimensionality of the data, i is 1, …, n, the input space is mapped with a mapping function Φ: rp→ F, X → X, wherein,
Figure BDA0001816224940000031
the covariance matrix C in the feature space is:
Figure BDA0001816224940000032
solving the eigenvalue lambda and eigenvector v of the covariance matrix C in the eigenspace, lambda being more than or equal to 0, and setting the eigenvalue of the covariance matrix C in the eigenspace to be more than or equal to 0 and less than or equal to lambda1≤λ2≤…≤λnCorresponding feature vector is v1，v2，…，νnIt is written as:
Figure BDA0001816224940000033
substituting formula (1) and formula (2) into middle phi (x)i)·Cv＝λ(Φ(xi) V) and let K ═ K (K)ij)n×n＝(Φ(xi)·Φ(xj) (i, j ═ 1, 2, …, n) to obtain K α ═ n λ α, where K is the kernel matrix and the eigenvalue of K is n λ αiThe feature vector of K is alpha1，α2，…，αnTaking the normalized eigenvector alpha corresponding to the first m eigenvalues1，α2，…，αmWherein, in the step (A),
Figure BDA0001816224940000034
Figure BDA0001816224940000035
Figure BDA0001816224940000036
at vrProjection gr(xj) Comprises the following steps:
Figure BDA0001816224940000037
is provided with
Figure BDA0001816224940000038
Then there is
Figure BDA0001816224940000041
Selecting the first M components in the formula (3) as principal components, reducing the dimension by KPCA to obtain an extracted principal component coefficient matrix alpha, and then performing linear transformation on the training sample X by using the extracted principal component coefficient matrix alpha to obtain the training sample X
Figure BDA0001816224940000042
And finally, training the corrosion prediction model of the generalized regression neural network by using the training sample X after linear transformation.
The generalized regression neural network is composed of an input layer, a hidden layer and a linear output layer.
The detection data of the corrosion outside the buried pipeline comprise soil resistivity, oxidation-reduction potential, chloride ion content, sulfate ion content, water content, PH, salt content, stray current, density of damage points, cathodic protection rate, natural potential and sulfide content at different detection points.
The invention has the following beneficial effects:
the invention relates to a buried pipeline external corrosion rate prediction method based on KPCA-BAS-GRNN, which is characterized in that during specific operation, a longicorn stigma search algorithm is adopted to optimize a smooth factor sigma of a corrosion prediction model, the influence of human factors during parameter selection is reduced, a test sample is input into the corrosion prediction model of a trained generalized regression neural network to obtain a predicted value, when the error between the predicted value and an actual measurement value is less than or equal to a preset value, the corrosion prediction model of the trained generalized regression neural network is correct, during actual prediction, detection data of the buried pipeline external corrosion to be predicted can be input into the trained corrosion prediction model to obtain the buried pipeline external corrosion rate, the reliability is high, and the longicorn stigma search (BAS) algorithm is an intelligent optimization algorithm developed based on the foraging principle of the longicorn in 2017, and adjusting the position of the longicorn according to comparison of the left and right longicorn odor intensity, and finally obtaining an optimal output value.
Drawings
FIG. 1 is a block diagram of a generalized recurrent neural network;
FIG. 2 is a schematic diagram of a simplified model of a longicorn;
FIG. 3 is a flow chart of the present invention;
FIG. 4 is a schematic diagram of an external corrosion index system for buried pipelines in a simulation experiment;
FIG. 5 is a graph of the effect of GRNN network training errors in a simulation experiment;
FIG. 6 is a graph of the effect of GRNN network prediction error in a simulation experiment;
FIG. 7 is a diagram of an iterative process of the BAS-GRNN model in a simulation experiment;
FIG. 8 is a comparison of different model test results in a simulation experiment;
FIG. 9 is a comparison graph of residuals of different model predictors in a simulation experiment.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
referring to fig. 1, the method for predicting the corrosion rate outside the buried pipeline based on KPCA-BAS-GRNN in the invention comprises the following steps:
1) acquiring detection data of the corrosion outside the buried pipeline, wherein the detection data of the corrosion outside the buried pipeline comprises soil resistivity, oxidation-reduction potential, chloride ion content, sulfate ion content, water content, PH, salt content, stray current, damage point density, cathodic protection rate, natural potential and sulfide content at different detection points, and then constructing an external corrosion index system of the pipeline according to the detection data of the corrosion outside the buried pipeline;
2) constructing a corrosion prediction model of the generalized regression neural network; the generalized regression neural network consists of an input layer, a hidden layer and a linear output layer, wherein in the hidden layer, the number of neurons is equal to the number of learning samples, and the weight LW of the network input and the first layer is calculated by using a dist distance formula1,1The node function of Q neurons in the layer is a Gaussian function,
Figure BDA0001816224940000061
n is the number of samples, sigma is a smooth factor, a linear function is adopted as a transfer function of the output layer, and finally, the predicted value expression output by the GRNN network is as follows: a is2＝purelin(LW2,1×a1/suma1)。
3) Dividing the data of the pipeline external corrosion index system obtained in the step 1) into a training sample and a test sample, training a corrosion prediction model of the generalized regression neural network through the training sample, optimizing a smooth factor sigma of the corrosion prediction model by adopting a Tianniu Lexus search algorithm, inputting the test sample into the trained corrosion prediction model of the generalized regression neural network to obtain a predicted value, judging the superiority and inferiority of the trained corrosion prediction model of the generalized regression neural network according to the predicted value and an actual measurement value, inputting the detection data of the to-be-predicted buried pipeline external corrosion into the trained corrosion prediction model of the generalized regression neural network to obtain the buried pipeline external corrosion rate when the trained corrosion prediction model of the generalized regression neural network is superior, and when the trained corrosion prediction model of the generalized regression neural network is inferior, go to step 1).
Specifically, let training sample X ═ X1，x2，…，xn}，xi∈Rp，RpFor the input space, P is the dimensionality of the data, i is 1, …, n, the input space is mapped with a mapping function Φ: rp→ F, X → X, wherein,
Figure BDA0001816224940000062
the covariance matrix C in the feature space is:
Figure BDA0001816224940000063
solving the eigenvalue lambda and eigenvector v of the covariance matrix C in the eigenspace, lambda being more than or equal to 0, and setting the eigenvalue of the covariance matrix C in the eigenspace to be more than or equal to 0 and less than or equal to lambda1≤λ2≤…≤λnCorresponding feature vector is v1，v2，…，νnIt is written as:
Figure BDA0001816224940000071
substituting formula (1) and formula (2) into middle phi (x)i)·Cv＝λ(Φ(xi) V) and let K ═ K (K)ij)n×n＝(Φ(xi)·Φ(xj) (i, j ═ 1, 2, …, n) to obtain K α ═ n λ α, where K is the kernel matrix and the eigenvalue of K is n λ αiThe feature vector of K is alpha1，α2，…，αnTaking the normalized eigenvector alpha corresponding to the first m eigenvalues1，α2，…，αmWherein, in the step (A),
Figure BDA0001816224940000072
Figure BDA0001816224940000073
Figure BDA0001816224940000074
at vrProjection gr(xj) Comprises the following steps:
Figure BDA0001816224940000075
is provided with
Figure BDA0001816224940000076
Then there is
Figure BDA0001816224940000077
Selecting the first M components in the formula (3) as principal components, reducing the dimension by KPCA to obtain an extracted principal component coefficient matrix alpha, and then performing linear transformation on the training sample X by using the extracted principal component coefficient matrix alpha to obtain the training sample X
Figure BDA0001816224940000078
And finally, training the corrosion prediction model of the generalized regression neural network by using the training sample X after linear transformation, wherein the specific flow is shown in FIG. 3, and the mathematical formula is described as follows:
a) creating random vector of longicorn beard orientation and carrying out normalization processing
Figure BDA0001816224940000081
Where rands () is a random function and k represents the spatial dimension.
b) Creating spatial coordinates of longicorn stigma left and right whiskers, i.e.
Figure BDA0001816224940000082
Wherein x isrtThe position coordinate of the t-th iteration of the right cow whisker is obtained; x is the number ofltThe position coordinate of the t-th iteration of the left antenna of the longicorn; x is the number oftThe centroid coordinate of the longicorn at the t iteration; d0Is the distance between two whiskers;
c) judging the left and right beard odor intensity according to the fitness function, namely f (x)l) And f (x)r) (x) is a fitness function;
d) iterative updating of longicorn positions
Figure BDA0001816224940000083
Wherein, deltatSign () is a sign function for the step factor of the t-th iteration.
The method selects the average relative error as a fitness function of the GRNN model, and optimizes sigma by adopting BAS.
In addition, the present invention employs an Average Relative Error (ARE) and a coefficient of determination (R)2) Evaluating the advantages and disadvantages of the corrosion prediction model, specifically:
Figure BDA0001816224940000084
Figure BDA0001816224940000091
wherein n is the number of samples,
Figure BDA0001816224940000092
is the predicted value of the ith sample, yiDetermining the value interval of the coefficient as [0,1 ] for the actual value of the ith sample]The closer the value is to 1, the better the model performance is, and the smaller the corresponding error is.
Simulation experiment
The east transportation of Sichuan gas is another long-distance pipeline network transportation project of natural gas after east transportation project of West gas in China is put into use, and the designed transportation capacity of pipelines is 120 hundred million m3And a, designing gas transmission pressure to be 10.0Mpa, pipe diameter to be 1016mm, steel pipe material to be X70, total length to be 2170Km, complicated soil along the line, multiple corrosion factors and serious corrosion condition outside the pipeline, and dividing the corrosion factors into 3 types of physics, chemistry and electrochemistry according to properties. The experiment adopts a real-time buried chip method, the corrosion rate is calculated through the mass change of the test chip, and no corrosion protection measure is set on the surface of the test chip. Along the lineSelecting 60 groups of test pieces in the monitoring points, recording the appearance of the test pieces, the description of corrosion products and soil parameters on site, and collecting soil samples. 60 groups of data are obtained through experimental analysis, and 12 main corrosion factors are determined to be used as an external corrosion index system, which is shown in figure 4. Finally, representative 15 sets of data were selected for predictive analysis, as detailed in table 1.
TABLE 1
Figure BDA0001816224940000093
Figure BDA0001816224940000101
Because the data dimensions of the external corrosion influence factors of the buried pipeline are different, firstly, the data are standardized, the result is shown in table 2, generally, KPCA has no fixed form requirement on the determination of the kernel function, and the invention selects RBF function: k (x, y) ═ exp (— | | x-y | | | non-conducting phosphor2/σ2) The normalized 12 variables were subjected to nonlinear feature extraction, and the calculated cumulative contribution ratios are shown in table 3.
TABLE 2
Figure BDA0001816224940000102
TABLE 3
Figure BDA0001816224940000103
As can be seen from Table 3, the cumulative contribution rate of the first 3 items exceeds 85%, so the first 3 main components are selected as the influence factors of the corrosion outside the buried pipeline, and the first 3 main components are F1，F2，F3Then the expression is as follows:
Figure BDA0001816224940000111
the prediction effect of the GRNN model is analyzed, 0.1, 0.2, 0.3, 0.4 and 0.5 are respectively selected as the spread values, 10 groups of training values and 5 groups of prediction values are constructed, and reference is made to fig. 5 and 6. The result shows that the prediction result of the GRNN model is unstable, accurate prediction of the corrosion rate of the pipeline cannot be realized, and the BAS algorithm is adopted to search for the optimal spread value, so that the stability and the accuracy of GRNN network prediction can be effectively improved.
From equation (10), 15 sets of data for KPCA reconstruction are available, setting BAS parameters: initial step size delta0And 3, dividing the extracted 15 groups of data into a training sample set and a testing sample set, and respectively substituting the training sample set and the testing sample set into a BAS-GRNN model for training and predicting, wherein the minimum RMSE after BAS optimization is 0.02042, the convergence state is reached after 54 times of iteration, and the spread value is 0.6747 according to the result shown in FIG. 7.
In order to verify the accuracy of the KPCA-BAS-GRNN model prediction, the output result is subjected to inverse normalization processing and is compared with a BP network and a PCA-SVM model, the training results of the three models are shown in figure 8, the difference between the actual value and the predicted value is represented by a residual error in order to visually express the model prediction effect, the result is shown in figure 9, and the comparison of the prediction results and the model performance evaluation of the three models are shown in table 4.
TABLE 4
Figure BDA0001816224940000112
Figure BDA0001816224940000121
As can be seen from Table 4, the average relative errors of BP, PCA-SVM and the KPCA-BAS-GRNN model are respectively 9.84%, 8.09% and 5.21%, the BAS-GRNN fitting result is better, the determining coefficient of the KPCA-BAS-GRNN model reaches 0.93, the model performance is greatly improved, and the prediction effect is better.
The matters not described in detail in the present specification belong to the prior art known to those skilled in the art, and the above embodiments are only for illustrating the present invention and not for limiting the present invention. Although the related embodiments of the present invention and the accompanying drawings are disclosed for illustrative purposes, those skilled in the art will appreciate that; various substitutions, changes, modifications and the like are possible without departing from the spirit and scope of the present invention and the appended claims. Therefore, all equivalent technical solutions also belong to the scope of the present invention, and the protection scope of the present invention should be defined by the claims, not limited to the disclosure of the best embodiment and the accompanying drawings.

Claims (4)

1. A KPCA-BAS-GRNN-based buried pipeline external corrosion rate prediction method is characterized by comprising the following steps:
1) acquiring detection data of corrosion outside the buried pipeline, and constructing an external pipeline corrosion index system according to the detection data of the corrosion outside the buried pipeline;
2) constructing a corrosion prediction model of the generalized regression neural network;
3) dividing the data of the pipeline external corrosion index system obtained in the step 1) into a training sample and a test sample, training a corrosion prediction model of the generalized regression neural network through the training sample, optimizing a smooth factor sigma of the corrosion prediction model by adopting a Tianniu Lexus search algorithm, inputting the test sample into the trained corrosion prediction model of the generalized regression neural network to obtain a predicted value, judging the superiority and inferiority of the trained corrosion prediction model of the generalized regression neural network according to the predicted value and an actual measurement value, inputting the detection data of the external corrosion of the buried pipeline to be predicted into the trained corrosion prediction model of the generalized regression neural network to obtain the external corrosion rate of the buried pipeline when the trained corrosion prediction model of the generalized regression neural network is superior, and when the trained corrosion prediction model of the generalized regression neural network is inferior, go to step 1).
2. The KPCA-BAS-GRNN based buried pipeline external corrosion rate prediction method according to claim 1,
let training sample X ═ X1，x2，…，xn}，xi∈Rp，RpFor the input space, P is the dimensionality of the data, i is 1, …, n, the input space is mapped with a mapping function Φ: rp→ F, X → X, wherein,
Figure FDA0003316954200000011
the covariance matrix C in the feature space is:
Figure FDA0003316954200000012
solving the eigenvalue lambda and eigenvector v of the covariance matrix C in the eigenspace, lambda being more than or equal to 0, and setting the eigenvalue of the covariance matrix C in the eigenspace to be more than or equal to 0 and less than or equal to lambda1≤λ2≤…≤λnCorresponding feature vector is v1，v2，…，νnIt is written as:
Figure FDA0003316954200000021
substituting the formula (1) and the formula (2) into phi (x)i)·Cv＝λ(Φ(xi) V) of phi (x)i) For a mapping function for converting an original data space into a high-dimensional data feature space, λ is an eigenvalue, v is an eigenvector, C is a covariance matrix, and let K ═ K (K)ij)n×n＝(Φ(xi)·Φ(xj) (i, j ═ 1, 2, …, n) to obtain K α ═ n λ α, where K is the kernel matrix and the eigenvalue of K is n λ αiThe feature vector of K is alpha1，α2，…，αnTaking the normalized eigenvector alpha corresponding to the first m eigenvalues1，α2，…，αmWherein, in the step (A),
Figure FDA0003316954200000022
r＝1,2,3，....，m，j＝1,2,3，....，m，
Figure FDA0003316954200000023
at vrProjection gr(xj) Comprises the following steps:
Figure FDA0003316954200000024
is provided with
Figure FDA0003316954200000025
Then there is
Figure FDA0003316954200000026
Wherein, KijFor the kernel function, i, j denote the rows and columns of samples, n x m dimensional samples;
selecting the first M components in the formula (3) as principal components, reducing the dimension by KPCA to obtain an extracted principal component coefficient matrix alpha, and then performing linear transformation on the training sample X by using the extracted principal component coefficient matrix alpha to obtain the training sample X
Figure FDA0003316954200000027
And finally, training the corrosion prediction model of the generalized regression neural network by using the training sample X after linear transformation.
3. The KPCA-BAS-GRNN-based buried pipeline external corrosion rate prediction method of claim 1, wherein the generalized regression neural network consists of an input layer, a hidden layer, and a linear output layer.
4. The KPCA-BAS-GRNN-based buried pipeline external corrosion rate prediction method of claim 1, wherein the detection data of the buried pipeline external corrosion comprises soil resistivity, oxidation-reduction potential, chloride ion content, sulfate ion content, water content, PH, salt content, stray current, breakage point density, cathodic protection rate, natural potential and sulfide content at different detection points.
CN201811143056.7A 2018-09-28 2018-09-28 A prediction method of external corrosion rate of buried pipeline based on KPCA-BAS-GRNN Expired - Fee Related CN109255490B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811143056.7A CN109255490B (en) 2018-09-28 2018-09-28 A prediction method of external corrosion rate of buried pipeline based on KPCA-BAS-GRNN

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811143056.7A CN109255490B (en) 2018-09-28 2018-09-28 A prediction method of external corrosion rate of buried pipeline based on KPCA-BAS-GRNN

Publications (2)

Publication Number Publication Date
CN109255490A CN109255490A (en) 2019-01-22
CN109255490B true CN109255490B (en) 2022-03-22

Family

ID=65048604

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811143056.7A Expired - Fee Related CN109255490B (en) 2018-09-28 2018-09-28 A prediction method of external corrosion rate of buried pipeline based on KPCA-BAS-GRNN

Country Status (1)

Country Link
CN (1) CN109255490B (en)

Families Citing this family (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102394826B1 (en) * 2018-02-14 2022-05-04 주식회사 엘지에너지솔루션 Power supply circuit for energy transfer between battery and smoothing capacitor and battery management system including the same
CN111668926B (en) * 2019-03-06 2023-09-01 中国电器科学研究院股份有限公司 Method for monitoring service microenvironment of distribution network equipment ring network unit in hot and humid climate
CN110087206B (en) * 2019-04-26 2021-12-21 南昌航空大学 Method for evaluating link quality by adopting generalized regression neural network
CN110138766B (en) * 2019-05-10 2020-06-12 福州大学 Network intrusion detection method combining longicorn stigma with random forest
CN110751339A (en) * 2019-10-24 2020-02-04 北京化工大学 Method, device and computer equipment for predicting corrosion rate of pipeline
CN111008648B (en) * 2019-11-12 2023-10-27 国网湖南省电力有限公司 Fault identification method, system and medium for pure optical fiber electronic current transformer
CN110852902A (en) * 2019-11-14 2020-02-28 合肥工业大学 Photovoltaic power generation power prediction method based on BAS-BP
CN111091149B (en) * 2019-12-16 2021-07-30 北京讯腾智慧科技股份有限公司 Gas leakage detection method, system, terminal and computer storage medium based on multi-source data fusion
CN111612733A (en) * 2020-04-02 2020-09-01 杭州电子科技大学 A convolutional neural network optimization method for medical image data analysis
CN112326540A (en) * 2020-09-30 2021-02-05 东莞材料基因高等理工研究院 Nuclear zirconium-4 alloy corrosion resistance prediction method based on BP network model, electronic equipment and storage medium
CN113094934B (en) * 2021-05-10 2023-02-17 重庆大学 Pipeline corrosion depth prediction method driven by physical model and data fusion
CN113361903B (en) * 2021-06-02 2023-12-12 中国石油大学(北京) Method, device and equipment for evaluating corrosion risk outside oil and gas pipeline
CN113268883B (en) * 2021-06-03 2024-06-04 西安建筑科技大学 Submarine crude oil pipeline corrosion rate prediction method based on PCA-ABC-SVM model
CN113239504B (en) * 2021-06-30 2022-01-28 西南石油大学 Pipeline corrosion defect prediction method based on optimized neural network
CN113419425A (en) * 2021-07-07 2021-09-21 哈尔滨理工大学 Intelligent regulation and control method for high-speed electric spindle water cooling system
CN113962266B (en) * 2021-10-25 2024-06-14 东北石油大学 Improved BAS-VMD-based pipeline leakage signal denoising method
CN114386462A (en) * 2022-01-10 2022-04-22 西安建筑科技大学 Corrosion rate prediction method, system, equipment and medium for submarine oil pipeline
CN115099151B (en) * 2022-07-01 2024-06-11 西南石油大学 Pipeline corrosion rate prediction method based on GWO-BP correction mechanism model
CN115078552B (en) * 2022-07-06 2023-09-08 江南大学 A method and system for defect detection of flip-chip soldering chips
CN117473302B (en) * 2023-12-26 2024-03-15 广东贝洛新材料科技有限公司 Method, device, equipment and storage medium for testing structural reliability of automobile parts

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6095426A (en) * 1997-11-07 2000-08-01 Siemens Building Technologies Room temperature control apparatus having feedforward and feedback control and method
CN103116759A (en) * 2013-02-01 2013-05-22 北京工业大学 Method applicable to identifying damage to space grid structure
CN105225010A (en) * 2015-10-12 2016-01-06 国网山东省电力公司电力科学研究院 A kind of transformer equipment lifetime estimation method based on reliability
CN107991873A (en) * 2017-12-11 2018-05-04 中国矿业大学 A kind of metro stray current leakage class Forecasting Methodology based on convolutional neural networks and BP neural network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6095426A (en) * 1997-11-07 2000-08-01 Siemens Building Technologies Room temperature control apparatus having feedforward and feedback control and method
CN103116759A (en) * 2013-02-01 2013-05-22 北京工业大学 Method applicable to identifying damage to space grid structure
CN105225010A (en) * 2015-10-12 2016-01-06 国网山东省电力公司电力科学研究院 A kind of transformer equipment lifetime estimation method based on reliability
CN107991873A (en) * 2017-12-11 2018-05-04 中国矿业大学 A kind of metro stray current leakage class Forecasting Methodology based on convolutional neural networks and BP neural network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
广义回归神经网络在燃气管道泄漏检测中的应用;郝永梅等;《消防管理研究》;20150715;950-953 *

Also Published As

Publication number Publication date
CN109255490A (en) 2019-01-22

Similar Documents

Publication Publication Date Title
CN109255490B (en) A prediction method of external corrosion rate of buried pipeline based on KPCA-BAS-GRNN
Seghier et al. Advanced intelligence frameworks for predicting maximum pitting corrosion depth in oil and gas pipelines
CN111291937A (en) Method for predicting quality of treated sewage based on combination of support vector classification and GRU neural network
CN112733443B (en) Water supply network model parameter optimization checking method based on virtual monitoring points
CN116341901B (en) Integrated evaluation method for landslide surface domain-monomer hazard early warning
CN112733997B (en) Hydrological time series prediction optimization method based on WOA-LSTM-MC
CN111950140B (en) A Method for Analyzing Seepage Behavior of Dams Considering Multiple Uncertainties
CN109858755B (en) A method for evaluating the quality of water bodies
CN107610021A (en) The comprehensive analysis method of environmental variance spatial and temporal distributions
Du et al. Deeppipe: Theory-guided prediction method based automatic machine learning for maximum pitting corrosion depth of oil and gas pipeline
CN112862178A (en) Ocean station water level space-time prediction method and device based on deep learning
CN102183802A (en) Short-term climate forecast method based on Kalman filtering and evolution modeling
CN116738822A (en) Drainage pipeline maximum corrosion depth prediction method based on LightGBM
Miao et al. Corroded submarine pipeline degradation prediction based on theory-guided IMOSOA-EL model
CN103353295B (en) A kind of method of accurately predicting dam dam body vertical deformation amount
CN117076893A (en) Sound velocity distribution forecasting method based on long-term and short-term memory neural network
Wang et al. Review of evaluation on ecological carrying capacity: The progress and trend of methodology
CN105005661A (en) Transmission tower material actual strength calculation method with practical result
CN106485351A (en) A kind of marine environmental monitoring erect-position layout optimization method
Li et al. [Retracted] Prediction of Chemical Corrosion Rate and Remaining Life of Buried Oil and Gas Pipelines in Changqing Gas Field
CN118734258A (en) Medium- and long-term runoff prediction method, device and medium
CN109766990B (en) Submarine pipeline corrosion rate prediction method based on AGA-ELM hybrid algorithm
CN117150925A (en) An inverse analysis method of mechanical parameters of rock mass in high and steep slopes of hydropower projects
Hu et al. Construction and application of a novel grey relational analysis model considering factor coupling relationship
CN112598163B (en) Grounding grid trenchless corrosion prediction model based on comparison learning and measurement learning

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20220322

CF01 Termination of patent right due to non-payment of annual fee