A system and method for automatically identifying and detecting defects in anti-corrosion layers of buried gas pipelines

By establishing a corrosion model that considers environmental factors and using a variety of detection technologies to identify and evaluate the damage points of buried gas pipelines, the problems of low accuracy of corrosion models and insufficient prediction accuracy in the prior art are solved, and more accurate damage is achieved Point identification and early warning improve pipeline safety and maintenance efficiency.

CN118643397BActive Publication Date: 2024-11-05SPECIAL EQUIP SAFETY SUPERVISION INSPECTION INST OF JIANGSU PROVINCE
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411105102.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-13
Publication Date
2024-11-05
Estimated Expiration
2044-08-13

AI Technical Summary

Technical Problem

The existing technology fails to effectively consider environmental factors, resulting in low accuracy of the corrosion model of the anti-corrosion layer of buried gas pipelines, and failure to accurately predict future damage points, affecting the safe operation of the pipeline.

Method used

By establishing a corrosion model that takes into account environmental factors, damage degree and service life, using technologies such as Pearson method, neural network and SVM, combined with measurement methods such as four-electrode depth method, dense separation potential method and gradient method, suspicious damage points are identified. And predict future damage points risks through databases.

Benefits of technology

It improves the accuracy and early warning capability of defect detection of anti-corrosion layer of buried gas pipelines, ensures the safe operation of the pipeline, and improves the maintenance efficiency of administrators.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118643397B_ABST
    Figure CN118643397B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of comprehensive detection, and discloses a system and method for automatically identifying and detecting defects in the anti-corrosion layer of a buried gas pipeline. The present invention first establishes a corrosion model and checks suspected damaged points through the influence of environmental factors, damage degree and service life; then obtains suspected damaged points through the Pearson method combined with the detector measurement, and substitutes the results into the corrosion model to verify whether the suspected damaged points are damaged points; then evaluates the damaged points through a neural network, and uses the cuckoo algorithm to obtain the optimal output value of the neural network, which greatly speeds up the convergence speed and avoids local optimality; evaluates the corrosion level according to the environmental factors around the damaged points, the damage degree and service life, and verifies the correctness of the damaged points; then establishes a database to store the damaged point information, establishes an SVM prediction model to predict according to the database information, and the administrator relies on the risk probability of the damaged points and the suspected damaged points in the future for maintenance.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of comprehensive detection, and in particular to a system and method for automatically identifying and detecting defects in an anti-corrosion layer of a buried gas pipeline. Background Art

[0002] Town gas is one of the important infrastructures of the city, and the buried pipeline, as the main means of gas transportation, has also developed rapidly in recent years. my country's gas pipeline network has entered a stage of rapid development. The buried gas pipeline has been in a relatively complex external environment for a long time. Therefore, with the increasing service time of the buried gas pipeline, the outer anti-corrosion layer of the buried gas pipeline often corrodes, ages, and breaks, which will eventually cause corrosion and perforation of the pipeline itself, and then gas leakage accidents will occur, which not only threatens people's lives and property safety, but also brings serious social harm. Therefore, evaluating the anti-corrosion layer of the buried gas pipeline and accurately locating the defects is an effective means to ensure the safe operation of the buried gas pipeline.

[0003] The Chinese invention patent application with the publication number CN114091320B discloses a natural gas pipeline corrosion failure time prediction method and device, including the following steps: by establishing a pipeline corrosion failure equation, determining the pipeline parameters that affect the pipeline corrosion failure time; selecting environmental parameters that affect the pipeline corrosion failure time; screening the pipeline parameters and environmental parameters based on correlation; using the screened parameters as input variables and the corrosion failure time as output variables, constructing a prediction model, and using the trained model to predict the pipeline corrosion failure time. The present invention determines the pipeline parameters that affect the corrosion failure time by establishing a pipeline corrosion failure equation, adds the pipeline environmental parameters, and uses the parameters screened based on correlation as input variables of the prediction model, which can significantly improve the prediction accuracy of the prediction model.

[0004] The existing methods do not fully consider environmental factors, the model accuracy is low, and a complete pipeline database has not been established, resulting in low prediction accuracy for future suspected breakpoints. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a system and method for automatically identifying and detecting defects in the anti-corrosion layer of a buried gas pipeline, so as to overcome the above-mentioned technical problems existing in the prior related technology.

[0006] To solve the above technical problems, the present invention is implemented by the following technical solutions: a method for automatically identifying and detecting defects in the anti-corrosion layer of a buried gas pipeline, the method comprising the following steps:

[0007] S1. Obtain the route of the buried gas pipeline, establish a corrosion model of the anti-corrosion layer of the buried gas pipeline based on environmental factors, damage degree and service life, and obtain the corrosion degree grade of the anti-corrosion layer of the buried gas pipeline;

[0008] S2, inputting a detection signal to the buried gas pipeline, using the Pearson method to detect changes in the detection signal of the anti-corrosion layer of the buried gas pipeline, and finding suspected damage points;

[0009] S3, using the corrosion model of the anti-corrosion layer of the buried gas pipeline to verify the suspected damage point and obtain the damage point;

[0010] S4. Evaluate the corrosion resistance of the buried gas pipeline at the damaged point based on a neural network to obtain damaged point information;

[0011] S5. Establish a database to store damaged point information and non-damaged point information to obtain a pipeline database;

[0012] S6. Identify the damaged point information in the pipeline database through SVM, and predict the risk probability value of the damaged point in the future.

[0013] The invention firstly establishes a corrosion model, takes into account the influence of environmental factors, degree of damage and service life, uses a four-electrode depth sounding method to measure soil resistivity, uses a close-interval potential method to measure the natural potential of the pipe-ground, uses a gradient method to measure DC stray current, and uses a PCM detection method to measure the degree of pipeline damage, and then establishes a corrosion model; then the Pearson method is used in combination with a detector to measure the location of the suspected damaged point, and the environmental factors, degree of damage and service life of the suspected damaged point are substituted into the corrosion model to determine whether it is a damaged point; the corrosiveness of the damaged point is evaluated based on a neural network, and the cuckoo algorithm is used to obtain the optimal output value of the output layer of the neural network, and the environmental factors, degree of damage and service life are input into the trained neural network model to obtain an evaluation grade to verify the accuracy of the damaged point; then a database is established to store the damaged point information, and an SVM prediction model is established to make predictions based on the information in the database to achieve early warning of damage; finally, the administrator conducts maintenance based on the risk probability value of the damaged point and the suspected damaged point in the future.

[0014] Preferably, the environmental factors in step S1 include soil resistivity, natural potential of pipe-ground and DC stray current, and the corrosion model of the anti-corrosion layer of the buried gas pipeline is:

[0015] F = 0.71E1 + 0.08E2 - 6.1 × 10 -5 E3+0.001E4-0.04E5-1.15,

[0016] Among them, E1, E2, E3, E4 and E5 are the degree of damage, service life, soil resistivity, pipe-ground natural potential and DC stray current respectively; F represents the corrosion degree coefficient. If F∈(4,5], the corrosion degree is 5, which means very serious; if F∈(3,4], the corrosion degree is 4, which means serious; if F∈(2,3], the corrosion degree is 3, which means medium; if F∈(1,2], the corrosion degree is 2, which means slight; if F∈(0,1], the corrosion degree is 1, which means no corrosion.

[0017] Preferably, the specific steps of S2 are:

[0018] A detection signal is transmitted at one end of the buried gas pipeline, and a detection signal is received at the other end of the buried gas pipeline; when the received detection signal strength is less than the transmitted detection signal strength, there is a suspected damage point in the buried gas pipeline, otherwise there is no suspected damage point.

[0019] Preferably, S3 comprises the following steps:

[0020] S31, measure the AC voltage gradient generated above the suspected damage point, record the suspected damage point as g, locate the position of the AC voltage gradient, calculate the damage degree, service life, soil resistivity, pipe-to-ground natural potential and DC stray current at the position of the AC voltage gradient, and substitute them into the corrosion model of the buried gas pipeline anti-corrosion layer to obtain the corrosion degree grade;

[0021] S32. Take an area within ε meters before and after the suspected damaged point on the buried gas pipeline; set a first threshold value ω1. When the absolute value of the difference between the average corrosion level of random points in the area and the corrosion level of the suspected damaged point is less than or equal to ω1, and the corrosion level of the suspected damaged point is 3, 4 or 5, the suspected damaged point is a damaged point; otherwise, the suspected damaged point is not a damaged point.

[0022] Preferably, S4 comprises the following steps:

[0023] S41, setting the data set consisting of the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of the λ1 group as the sample index set of the input layer of the RBF neural network, normalizing the sample index set, inputting the normalized data set as the sample set into the RBF neural network, training the RBF neural network until the RBF neural network converges, and obtaining a trained RBF neural network model;

[0024] S42, setting the data set consisting of the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of group λ2 as the detection set, normalizing the detection set, and inputting the normalized detection set into the trained RBF neural network model;

[0025] When the output result meets the corrosion degree level of the corresponding buried gas pipeline anti-corrosion layer, the trained RBF neural network model is used as the final RBF neural network model;

[0026] S43. When the output result does not meet the corresponding corrosion level of the buried gas pipeline anti-corrosion layer, the initial weight set of the trained RBF neural network model is set to h = {h1, h2, ..., h n}, where n represents the number of weights; set the size of the cuckoo population to k, set the maximum number of iterations and the current number of iterations, and the initial cuckoo weight set corresponding to the kth cuckoo in the cuckoo population is p = {p k1 ,p k2 ,...,p kn}, where p kn represents the nth weight corresponding to the kth cuckoo;

[0027] The initial weight matrix is ​​used as the initial nest position matrix of the cuckoo population, the mean square error target value is set as the fitness value, and the weights corresponding to the cuckoo are iteratively updated according to the fitness value. The mean square error target value is The calculation formula is as follows:

[0028]

[0029] Where p represents the maximum number of iterations, q′ represents the output value, q″ represents the input value, and H represents the current number of iterations;

[0030] When the current number of iterations is greater than or equal to the maximum number of iterations, the iteration is stopped to obtain the final weight matrix;

[0031] Selecting the nest position corresponding to the minimum fitness from the final weight matrix as the optimal weight, substituting each position component in the optimal weight into the RBF neural network to obtain an optimized RBF neural network model;

[0032] S44, the optimized RBF neural network model is input into the optimized RBF neural network model using the normalized detection set in S42, and if the output result meets the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, the final RBF neural network model is obtained; if the output result does not meet the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, the weights of the RBF neural network model are continuously optimized using the method of S43 until the normalized detection set in S42 is input into the optimized RBF neural network model, and the output result meets the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, thereby obtaining the final RBF neural network model;

[0033] S45, setting the final RBF neural network model output evaluation level;

[0034] S46, inputting the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of the damaged point into the final RBF neural network model to obtain an evaluation grade.

[0035] Preferably, S6 comprises the following steps:

[0036] S61, constructing a trained SVM prediction model;

[0037] S62. Obtain the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of the new damaged point to form an SVM test set. Input the SVM test set into the trained SVM prediction model. The output layer of the trained SVM prediction model outputs a judgment result, where the output 0 indicates damage and 1 indicates no damage. Input the damaged point information and non-damaged point information in the pipeline database excluding the SVM training set and the SVM test set into the SVM prediction model to predict the risk probability value of the damaged point in the future.

[0038] Preferably, the S6 further comprises: inspecting and repairing the suspected damaged points whose risk probability values ​​of future damaged points exceed a set threshold and which are received by the receiver.

[0039] Preferably, the S61 specifically includes the following steps: collecting damaged point information and non-damaged point information in the pipeline database, selecting μ1 damaged points and μ1 non-damaged points as an SVM training set; then selecting μ2 damaged points and μ2 non-damaged points as an SVM test set; after normalizing the SVM training set and the SVM test set, inputting the SVM training set into a support vector classifier model for training until the support vector classifier model converges, stopping the training, and obtaining a trained SVM prediction model.

[0040] The invention uses the damage point information in the database to input the training set into the support vector classifier model training. The obtained model can predict the risk probability of damage points in the future. The administrator can perform early maintenance based on the risk probability of damage points in the future to prevent accidents caused by damage to the anti-corrosion layer. At the same time, it is convenient for the administrator to perform maintenance and improve work efficiency.

[0041] The present invention also discloses an automatic identification and detection system for anti-corrosion layer defects of buried gas pipelines, which specifically comprises: a corrosion model module, a damage point inspection and evaluation module, a database module, and a damage point prediction module;

[0042] The corrosion model module is used to establish a corrosion model using parameters and verify suspicious damage points;

[0043] The damage point inspection and evaluation module is used to find the damage point and evaluate the accuracy of the damage point;

[0044] The database module is used to store damage point information;

[0045] The damage point prediction module is used to predict the risk probability of the damage point using the SVM prediction model according to the damage point information in the database.

[0046] The present invention has the following beneficial effects:

[0047] 1. The invention firstly establishes a corrosion model, takes into account the influence of environmental factors, degree of damage and service life, uses a four-electrode depth sounding method to measure soil resistivity, uses a close-interval potential method to measure the natural potential of the pipe-ground, uses a gradient method to measure DC stray current, and uses a PCM detection method to measure the degree of pipeline damage, and then establishes a corrosion model; then the Pearson method is used in combination with a detector to measure the location of the suspected damaged point, and the environmental factors, degree of damage and service life of the suspected damaged point are substituted into the corrosion model to determine whether it is a damaged point; the corrosiveness of the damaged point is evaluated based on a neural network, and the cuckoo algorithm is used to obtain the optimal output value of the output layer of the neural network, and the environmental factors, degree of damage and service life are input into the trained neural network model to obtain an evaluation grade to verify the accuracy of the damaged point; then a database is established to store the damaged point information, and an SVM prediction model is established to predict the information in the database to achieve early warning of damage; finally, the administrator performs maintenance based on the risk probability of future damaged points and the suspected damaged points.

[0048] 2. The invention establishes a corrosion model for the anti-corrosion layer of the buried gas pipeline, takes into account multiple influencing factors such as degree of damage, service life, soil resistivity, natural potential of the pipe and ground, and DC stray current, and uses PCM detection method, four-electrode depth sounding method, close-interval potential method and gradient method for calculation and measurement. The obtained values ​​are fitted with software to establish a linear relationship, and the corrosion model for the anti-corrosion layer of the buried gas pipeline is obtained, which is beneficial to the subsequent verification of the damage point.

[0049] 3. The invention finds the optimal output value of the output layer of the RBF neural network through the cuckoo algorithm, wherein the location where the cuckoo searches for the nest is set, and the possibility of being discovered by the nest owner is taken into consideration. The cuckoo continuously iterates the location of the nest to obtain the optimal nest location, that is, the optimal output value, and the search efficiency is high; the optimal output value is input into the neural network, and the neural network is trained after normalization to obtain a trained neural network model. The output result verifies the output result of the corrosion model of the anti-corrosion layer of the buried gas pipeline. The method of use is rigorous and the result is accurate.

[0050] 4. The invention uses the damage point information in the database to input the training set into the support vector classifier model training. The obtained model can predict the risk probability of damage points in the future. The administrator can conduct advance inspections on the risk probability of damage points in the future to prevent accidents caused by damage to the anti-corrosion layer. At the same time, it is convenient for the administrator to conduct inspections and improves work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0052] Figure 1 A schematic flow chart of a method for automatically identifying and detecting defects in an anti-corrosion layer of a buried gas pipeline provided by the present invention. DETAILED DESCRIPTION

[0053] Combined with Figure 1 , the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0055] This embodiment discloses a method for automatically identifying and detecting defects in the anti-corrosion layer of a buried gas pipeline, which specifically includes the following contents:

[0056] S1. Obtain the route of the buried gas pipeline, establish a corrosion model of the anti-corrosion layer of the buried gas pipeline based on environmental factors, damage degree and service life, and obtain the corrosion degree grade of the anti-corrosion layer of the buried gas pipeline;

[0057] The S1 comprises the following steps:

[0058] S11, obtaining the route of the buried gas pipeline, inputting the route of the buried gas pipeline into the cloud, so that the operator can view the location and coordinate information of the route of the buried gas pipeline on the map in real time, and obtain the buried gas pipeline to be processed;

[0059] S12, considering the influence of environmental factors on the buried gas pipeline to be treated to establish a corrosion model of the anti-corrosion layer of the buried gas pipeline, the specific steps are as follows:

[0060] S121. The environmental factors include soil resistivity. The soil resistivity is measured using a four-electrode depth measurement method. When the distance between the measuring electrodes is greater than or equal to 20 times the depth of the measuring electrodes, the calculation formula is as follows:

[0061]

[0062] Wherein, soil resistivity is A, measured resistance is a, measuring electrode spacing is b, and measuring electrode depth is b′;

[0063] When the distance between the measuring electrodes is less than or equal to 0.2 times the depth of the measuring electrodes, the calculation formula is as follows:

[0064] A = 2πab;

[0065] S122, the environmental factors also include the natural potential of the pipe-ground. The natural potential of the pipe-ground is measured by using a close-interval potential method. An electrode is moved along the buried gas pipeline to be treated, and then a potential data collector is used to collect the natural potential of the pipe-ground at a specific interval. The data is input into a potential software to obtain a continuous natural potential curve of the pipe-ground. The continuous natural potential curve of the pipe-ground corresponds to the natural potential of the pipe-ground.

[0066] S123, the environmental factors also include DC stray current, the natural potential of the buried gas pipeline to be treated has a positive offset greater than or equal to 20 millivolts, and the buried gas pipeline to be treated has DC stray current; the DC stray current is measured using the gradient method, three electrodes are placed in the soil, recorded as the first electrode, the second electrode and the third electrode, the potential difference between the first electrode and the second electrode, and the second electrode and the third electrode is measured respectively, the potential difference is vector synthesized to obtain an electric field, and the potential gradient is calculated in the electric field, and the calculation formula is as follows:

[0067]

[0068] Among them, B represents the potential gradient, Δc represents the potential difference, and d represents the number of electrons gained and lost by the three electrodes;

[0069] The value corresponding to the potential gradient is the stray current;

[0070] S13. Considering the influence of pipeline damage degree on the buried gas pipeline to be treated, a corrosion model of the buried gas pipeline anti-corrosion layer is established. The pipeline damage degree is measured by using the PCM detection method. First, a signal current is passed into the pipeline, and then the pipeline parameters are calculated according to the change in magnetic field strength caused by the current change. The calculation formula is as follows:

[0071]

[0072] Wherein, e represents the pipeline diameter, α represents the attenuation coefficient of the current in the pipeline, and C represents the pipeline parameter. When C>10000, it means that the pipeline is not damaged; when 10000≥C≥5000, it means that the pipeline is slightly damaged; when C<5000, it means that the pipeline is moderately damaged; when 5000≥C≥3000, it means that the pipeline is severely damaged; when C<3000, it means that the pipeline is seriously damaged.

[0073] S14. Conduct relevant analysis on the environmental factors, damage degree and service life. The specific steps are as follows:

[0074] S141. The partial correlation coefficient is used to measure the linear relationship between environmental factors, damage degree and service life, which are set as variables D1, D2 and D3 respectively. The second-order partial correlation coefficient calculation formula of D1, D2 and D3 is as follows:

[0075]

[0076] Among them, f 12 Indicates the correlation coefficient between D1 and D2, f 13 Indicates the correlation coefficient between D1 and D3, f 23 represents the correlation coefficient between D2 and D3, and β represents the second-order partial correlation coefficient of D1, D2 and D3;

[0077] S142, verifying the second-order partial correlation coefficient, the verification formula is as follows:

[0078]

[0079] Among them, f 123 represents the correlation coefficient between D1, D2 and D3, g represents the number of related factors, when the verification result χ is 0, it means that D1, D2 and D3 have no significant relationship; when the verification result χ is 1, it means that D1, D2 and D3 have a significant relationship;

[0080] S143. χ=1 is obtained by calculation. There is a significant relationship between environmental factors, damage degree and service life. The weight coefficients of environmental factors, damage degree and service life are calculated by using the main factor screening method. The damage degree, service life, soil resistivity, pipe-to-ground natural potential and DC stray current are set to be E1, E2, E3, E4 and E5 respectively. E1, E2, E3, E4 and E5 are input into SPSS software. E1, E2, E3, E4 and E5 are linearly correlated.

[0081] S15. A corrosion model of the anti-corrosion layer of the buried gas pipeline is established according to the damage degree, service life, soil resistivity, natural potential of the pipe and the ground, and DC stray current. The calculation formula is as follows:

[0082] F = 0.71E1 + 0.08E2 - 6.1 × 10 -5 E3+0.001E4-0.04E5-1.15,

[0083] Wherein, F represents the corrosion degree coefficient. If F∈(4,5], the corrosion degree is 5, which means very serious. If F∈(3,4], the corrosion degree is 4, which means serious. If F∈(2,3], the corrosion degree is 3, which means medium. If F∈(1,2], the corrosion degree is 2, which means slight. If F∈(0,1], the corrosion degree is 1, which means no corrosion.

[0084] S2, inputting a detection signal to the buried gas pipeline, using the Pearson method to detect changes in the detection signal of the anti-corrosion layer of the buried gas pipeline, and finding suspected damage points;

[0085] The S2 comprises the following steps:

[0086] S21, inputting a detection signal at one end of the buried gas pipeline, the detection signal passes through the damaged point of the buried gas pipeline, and generates an AC voltage gradient above the damaged point, at which time the detection signal strength will decrease;

[0087] S22, receiving a detection signal at the other end of the buried gas pipeline, when the strength of the received detection signal is less than the strength of the transmitted detection signal, the buried gas pipeline has a suspected damage point, otherwise there is no suspected damage point;

[0088] S3, using the corrosion model of the anti-corrosion layer of the buried gas pipeline to verify the suspected damage point and obtain the damage point;

[0089] The S3 comprises the following steps:

[0090] S31, measuring the AC voltage gradient generated above the suspected damage point, recording the suspected damage point as g, locating the AC voltage gradient position, the operator goes to the position where the AC voltage gradient is generated, calculates the damage degree, service life, soil resistivity, pipe-to-ground natural potential and DC stray current at the position where the AC voltage gradient is generated; and substitutes them into the corrosion model of the buried gas pipeline anti-corrosion layer to obtain the corrosion degree grade;

[0091] S32, take the area within ε meters before and after the suspected damage point on the buried gas pipeline, and record it as the set G = (g1, g2, ..., g ε ), where g1,g2,...,g εRepresent random points in the area, and calculate the average value of the corrosion level of the random points in the area respectively; set a first threshold ω1, when the absolute value of the difference between the average value of the corrosion level of the random points in the area and the corrosion level of the suspected damaged point is less than or equal to ω1, and the corrosion level of the suspected damaged point is 3 or 4 or 5, the suspected damaged point is a damaged point; otherwise, the suspected damaged point is not a damaged point;

[0092] S4. Evaluate the corrosion resistance of the buried gas pipeline at the damaged point based on a neural network to obtain damaged point information;

[0093] The S4 comprises the following steps:

[0094] S41, setting the data set consisting of the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of the λ1 group as the sample index set of the input layer of the RBF neural network, normalizing the sample index set, inputting the normalized data set as the sample set into the RBF neural network, training the RBF neural network until convergence, and the output layer outputs the optimal output value; obtaining a trained RBF neural network model;

[0095] S42, setting the data set consisting of the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of group λ2 as the detection set, normalizing the detection set, and inputting the normalized detection set into the trained RBF neural network model,

[0096] When the output result meets the corrosion degree level of the corresponding buried gas pipeline anti-corrosion layer, the trained RBF neural network model is used as the final RBF neural network model;

[0097] S43. When the output result does not meet the corresponding corrosion level of the buried gas pipeline anti-corrosion layer, the initial weight set of the trained RBF neural network model is set to h = {h1, h2, ..., h n}, where n represents the number of weights; set the size of the cuckoo population to k, set the maximum number of iterations and the current number of iterations, and the initial cuckoo weight set corresponding to the kth cuckoo in the cuckoo population is p = {p k1 ,p k2 ,...,p kn}, where p kn represents the nth weight corresponding to the kth cuckoo, setting the weight interval to [p1, p2], and the step length to The weight distance is The initial weight set randomly generates an initial weight matrix p′ according to the weight distance as follows:

[0098]

[0099] The initial weight matrix is ​​used as the initial nest position matrix of the cuckoo population, the mean square error target value is set as the fitness value, and the weights corresponding to the cuckoo are iteratively updated according to the fitness value. The mean square error target value is The calculation formula is as follows:

[0100]

[0101] Where p represents the maximum number of iterations, q′ represents the output value, q″ represents the input value, and H represents the current number of iterations;

[0102] When the current number of iterations is greater than or equal to the maximum number of iterations, the iteration is stopped and the final weight matrix p″ is obtained as follows:

[0103]

[0104] Among them, p′ kn represents the nth final weight corresponding to the kth cuckoo;

[0105] Selecting the nest position corresponding to the minimum fitness from the final weight matrix as the optimal weight, substituting each position component in the optimal weight into the RBF neural network to obtain an optimized RBF neural network model;

[0106] S44, the optimized RBF neural network model is input into the optimized RBF neural network model using the normalized detection set in S42, and if the output result meets the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, the final RBF neural network model is obtained; if the output result does not meet the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, the weights of the RBF neural network model are continuously optimized using the method of S43 until the normalized detection set in S42 is input into the optimized RBF neural network model, and the output result meets the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, thereby obtaining the final RBF neural network model;

[0107] S45, setting the final RBF neural network model output evaluation level, the evaluation level is divided into 1, 2, 3, 4, 5 levels, respectively representing no corrosion, slight corrosion, moderate corrosion, severe corrosion, very severe corrosion;

[0108] S46, inputting the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of the damaged point into the final RBF neural network model to obtain an evaluation grade.

[0109] S5, establishing a database, storing the damage point information in the database, and obtaining a pipeline database;

[0110] The S5 comprises the following steps:

[0111] S51. Establish a static table in the database, wherein the static table includes buried gas pipeline information, environmental factor information, damage degree information, service life information, maintenance and renovation information, and data record information; the buried gas pipeline information includes pipeline number, pipeline length, pipeline diameter, and pipeline starting point and end point; the environmental factor information includes soil resistivity, pipe-ground natural potential, and DC stray current; the damage degree information includes corrosion degree grade; the service life information includes pipeline production date and pipeline commissioning date; the maintenance and renovation information includes pipeline number, maintenance and renovation type, maintenance and renovation date, and maintenance and renovation location; the inspection record information includes pipeline number, inspection date, and inspector name;

[0112] S52, establishing a database dynamic table, the dynamic table includes PCM (pulse code modulation) detection data and PCM transmission and reception data, the PCM detection data includes pipeline number, starting point and end point distance and burial depth; PCM transmission and reception data includes starting point and end point detection signals, damage point detection signals and inspector name;

[0113] S53, storing the damaged point information and the non-damaged point information in the database to obtain a pipeline database;

[0114] S6. Identify the damaged point information in the pipeline database through SVM and predict the risk probability of the damaged point in the future. The administrator inspects the suspected damaged points whose risk probability value of the damaged point in the future exceeds the set threshold and which are received by the receiver;

[0115] The S6 comprises the following steps:

[0116] S61, collecting damaged point information and non-damaged point information in the pipeline database, selecting μ1 damaged points and μ1 non-damaged points as SVM training sets; then selecting μ2 damaged points and μ2 non-damaged points as SVM test sets; after normalizing the SVM training set and the SVM test set, inputting the SVM training set into a support vector classifier model for training until convergence, stopping the training, and obtaining a trained SVM prediction model;

[0117] S62, input the SVM test set into the trained SVM prediction model, set the penalty factor and kernel parameter in the evaluation parameters, and after smoothing with the Gaussian function, the output layer of the trained SVM prediction model outputs the judgment result, where the output 0 indicates damage and 1 indicates no damage; input the damaged point information and non-damaged point information in the pipeline database excluding the SVM training set and the SVM test set into the SVM prediction model, predict the risk probability of damaged points in the future, set the prediction threshold ω2, if the risk probability of damaged points in the future is greater than or equal to ω2, the administrator will inspect the location where the risk probability of damaged points in the future exceeds the set threshold; otherwise, the administrator will not inspect;

[0118] S63, the administrator performs maintenance based on the risk probability of future damage points and the suspected damage points received by the receiver; if it is a damage point, dig up the damage point for repair; if it is not a damage point, upload it to the pipeline database.

[0119] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0120] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A method for automatic identification and detection of anti-corrosion layer defects of buried gas pipelines. It is characterized in that The steps include: S1. Obtain the route of the buried gas pipeline, establish a corrosion model of the anti-corrosion layer of the buried gas pipeline based on environmental factors, damage degree and service life, and obtain the corrosion degree grade of the anti-corrosion layer of the buried gas pipeline; S2, inputting a detection signal to the buried gas pipeline, using the Pearson method to detect changes in the detection signal of the anti-corrosion layer of the buried gas pipeline, and finding suspected damage points; S3, using the corrosion model of the anti-corrosion layer of the buried gas pipeline to verify the suspected damage point and obtain the damage point; The S3 comprises the following steps: S31, measure the AC voltage gradient generated above the suspected damage point, record the suspected damage point as g, locate the position of the AC voltage gradient, calculate the damage degree, service life, soil resistivity, pipe-to-ground natural potential and DC stray current at the position of the AC voltage gradient, and substitute them into the corrosion model of the buried gas pipeline anti-corrosion layer to obtain the corrosion degree grade; S32, select the area within ε meters before and after the suspected damage point on the buried gas pipeline; set the first threshold ω 1 , when the absolute value of the difference between the average corrosion level of random points in the area and the corrosion level of the suspected damage point is less than or equal to ω 1 When the suspected damaged point has a corrosion degree of 3, 4 or 5, the suspected damaged point is a damaged point; otherwise, the suspected damaged point is not a damaged point; S4. Evaluate the corrosion resistance of the buried gas pipeline at the damaged point based on a neural network to obtain damaged point information; The S4 comprises the following steps: S41. Setting λ 1 The data set consisting of the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current is used as a sample index set for inputting the RBF neural network input layer, the sample index set is normalized, the normalized data set is used as a sample set for inputting the RBF neural network, the RBF neural network is trained until the RBF neural network converges, and a trained RBF neural network model is obtained; S42, setting λ 2 A data set consisting of damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current is used as a detection set, the detection set is normalized, and the normalized detection set is input into the trained RBF neural network model; When the output result meets the corrosion degree level of the corresponding buried gas pipeline anti-corrosion layer, the trained RBF neural network model is used as the final RBF neural network model; S43. When the output result does not meet the corrosion level of the corresponding buried gas pipeline anti-corrosion layer, the initial weight set of the trained RBF neural network model is set to h = {h 1 ,h 2 ,...,h n }, where n represents the number of weights; set the size of the cuckoo population to k, set the maximum number of iterations and the current number of iterations, and the initial cuckoo weight set corresponding to the kth cuckoo in the cuckoo population is p = {p k1 ,p k2 ,...,p kn }, where p kn represents the nth weight corresponding to the kth cuckoo; The initial weight matrix is ​​used as the initial nest position matrix of the cuckoo population, the mean square error target value is set as the fitness value, and the weights corresponding to the cuckoo are iteratively updated according to the fitness value. The mean square error target value is The calculation formula is as follows: Where p represents the maximum number of iterations, q′ represents the output value, q″ represents the input value, and H represents the current number of iterations; When the current number of iterations is greater than or equal to the maximum number of iterations, the iteration is stopped to obtain the final weight matrix; Selecting the nest position corresponding to the minimum fitness from the final weight matrix as the optimal weight, substituting each position component in the optimal weight into the RBF neural network to obtain an optimized RBF neural network model; S44, the optimized RBF neural network model is input into the optimized RBF neural network model using the normalized detection set in S42, and if the output result meets the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, the final RBF neural network model is obtained; if the output result does not meet the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, the weights of the RBF neural network model are continuously optimized using the method of S43 until the normalized detection set in S42 is input into the optimized RBF neural network model, and the output result meets the corresponding corrosion degree level of the anti-corrosion layer of the buried gas pipeline, thereby obtaining the final RBF neural network model; S45, setting the final RBF neural network model output evaluation level; S46, inputting the damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current of the damaged point into the final RBF neural network model to obtain an evaluation grade; S5. Establish a database to store damaged point information and non-damaged point information to obtain a pipeline database; S6. Identify the damaged point information in the pipeline database through SVM, and predict the risk probability value of the damaged point in the future; The S6 comprises the following steps: S61, constructing a trained SVM prediction model; S62, obtaining the damage degree, service life, soil resistivity, natural potential of the pipe and ground, and DC stray current of the new damaged point to form an SVM test set, inputting the SVM test set into the trained SVM prediction model, and the output layer of the trained SVM prediction model outputs a judgment result, wherein the output 0 indicates damage and 1 indicates no damage; inputting the damaged point information and the non-damaged point information in the pipeline database excluding the SVM training set and the SVM test set into the SVM prediction model to predict the risk probability value of the damaged point in the future; The S6 further includes: inspecting and repairing the suspected damaged points whose risk probability value of the future damaged points exceeds the set threshold and which are received by the receiver; The S61 specifically includes the following steps: Collect the damaged point information and non-damaged point information in the pipeline database, select μ 1 Damage points, μ 1 undamaged points as the SVM training set; then select μ 2 Damage points, μ 2 undamaged points as the SVM test set; after the SVM training set and the SVM test set are normalized, the SVM training set is input into the support vector classifier model for training until the support vector classifier model converges, the training is stopped, and a trained SVM prediction model is obtained.

2. According to claim 1, a method for automatically identifying and detecting defects in the anti-corrosion layer of a buried gas pipeline, It is characterized in that The environmental factors in step S1 include soil resistivity, natural potential of pipe-ground and DC stray current. The corrosion model of the anti-corrosion layer of the buried gas pipeline is: F=0.71E 1 +0.08E 2 -6.1×10 -5 AND 3 +0.001E 4 -0.04E 5 -1.15; Among them, E 1 、E 2 、E 3 、E 4 and E 5 They are damage degree, service life, soil resistivity, pipe-ground natural potential and DC stray current respectively; F represents the corrosion degree coefficient. If F∈(4,5], the corrosion degree is 5, which means very serious; if F∈(3,4], the corrosion degree is 4, which means serious; if F∈(2,3], the corrosion degree is 3, which means medium; if F∈(1,2], the corrosion degree is 2, which means slight; if F∈(0,1], the corrosion degree is 1, which means no corrosion.

3. According to claim 2, a method for automatically identifying and detecting defects in the anti-corrosion layer of a buried gas pipeline, It is characterized in that The specific steps of S2 are: A detection signal is transmitted at one end of the buried gas pipeline, and a detection signal is received at the other end of the buried gas pipeline; when the received detection signal strength is less than the transmitted detection signal strength, there is a suspected damage point in the buried gas pipeline, otherwise there is no suspected damage point.

4. A system for realizing the method for automatically identifying and detecting defects of anti-corrosion layer of buried gas pipelines as claimed in any one of claims 1 to 3, It is characterized in that Specifically include: Corrosion model module, damage point inspection and evaluation module, database module, damage point prediction module; The corrosion model module is used to establish a corrosion model using parameters and verify suspicious damage points; The damage point inspection and evaluation module is used to find the damage point and evaluate the accuracy of the damage point; The database module is used to store damage point information; The damage point prediction module is used to predict the risk probability of the damage point using the SVM prediction model according to the damage point information in the database.