Pipeline defect growth prediction method and system based on multiple rounds of internal detection data

Abstract

The invention provides a pipeline defect growth prediction method based on multi-round internal detection data, which dynamically predicts pipeline defect growth based on the multi-round internal detection data and a Markov chain, and comprises the following steps: acquiring M-round internal detection data; performing data alignment on the detection data in the M rounds based on a data alignment algorithm; newly added defects relative to the previous round are identified based on the aligned multiple rounds of inner detection data; on the basis of a coincidence coefficient algorithm and the aligned multi-round internal detection data, carrying out growth defect analysis on the newly added defects, and identifying that the growth defects of the pipeline in the current round are active corrosion or inactive corrosion relative to the growth defects in the previous round during detection; a Markov chain state transition matrix is constructed based on M rounds of internal detection data of the pipeline, so that the defect growth condition of the pipeline during the next round of internal detection is intelligently predicted; and carrying out pipeline defect intrinsic safety evaluation based on an intelligent prediction result, and carrying out corresponding maintenance decision making. The invention further discloses a corresponding system, electronic equipment and a computer readable storage medium.

Description

Technical Field

[0001] This invention relates to the field of pipeline safety evaluation technology, specifically to a method and system for predicting pipeline defect growth based on multi-round inspection data. Background Technology

[0002] In recent years, China's natural gas pipeline network has developed rapidly. By 2025, the total length of the national natural gas pipeline network is expected to reach 163,000 kilometers, covering cities with a population of over 500,000, and gradually forming a nationwide basic network of "interconnected trunk lines and regional networks." However, the expansion of the pipeline network has also been accompanied by the accumulation of safety risks. According to statistics, corrosion, construction defects, and third-party sabotage are the main causes of gas pipeline accidents both domestically and internationally, accounting for more than 60%. Therefore, strengthening the management of natural gas pipeline defects and accurately assessing the activity status of pipeline defects are urgent needs for achieving healthy management of natural gas pipelines.

[0003] Among existing pipeline defect analysis methods, model-based corrosion rate prediction methods are widely used, but this method has the following drawbacks:

[0004] (1) Difficulty in model construction: Pipeline corrosion is affected by a variety of environmental factors, such as soil properties, internal media, and operating conditions. The corrosion rate generally exhibits nonlinear characteristics. Therefore, it is difficult to construct a high-precision corrosion rate prediction model, as it is hard to comprehensively consider multiple factors.

[0005] (2) Difficulty in model application: The application of corrosion rate prediction models usually requires obtaining the initial state information of defects. However, due to the influence of the location environment of the pipeline and the surrounding buildings, not all defects are suitable for excavation. Therefore, it is difficult to obtain the initial state data of pipeline defects.

[0006] (3) Data fragmentation: Multiple rounds of detection data are not effectively correlated, making it difficult to capture the evolution pattern of defects.

[0007] In summary, traditional defect evaluation methods are mostly based on deterministic analysis (such as residual strength evaluation and linear fitting of corrosion rates). Corrosion rate prediction methods generally require determining the initial state of the defect, such as the corrosion depth and width. However, in practical field applications, it is found that many defects emerge after a round of inspections, and each defect, influenced by the pipeline's geographical location and surrounding buildings, does not necessarily meet the conditions for on-site excavation. Therefore, model-based corrosion rate prediction methods are not applicable to all defects. Among numerous pipeline defects, failure to promptly predict and maintain active defects can lead to pipeline corrosion and leakage, causing incalculable property damage and even personal injury. Therefore, there is an urgent need for a natural gas pipeline defect analysis method that can improve the intrinsic safety analysis level of natural gas pipeline defects for enterprises while minimizing excavation costs. Summary of the Invention

[0008] To overcome the deficiencies and shortcomings in existing technologies, this invention provides a pipeline defect growth prediction method and system based on multi-round inspection data. This statistically-based defect growth prediction method constructs a defect state transition matrix by statistically analyzing the defect growth status of pipeline inspection data across three rounds. This quantifies the randomness of defect evolution and predicts the growth status of defects in the next round. Defects predicted to grow are given priority attention, while those predicted not to grow are given less attention. This achieves high-precision predictive maintenance decision support, applicable to complex scenarios such as long-distance pipelines and subsea pipelines. The proposed method can significantly reduce the economic burden on enterprises in pipeline intrinsic safety analysis and provide data support for pipeline operation and maintenance management.

[0009] The first aspect of this invention is to provide a pipeline defect growth prediction method based on multi-round intra-inspection data, which dynamically predicts pipeline defect growth based on multi-round intra-inspection data and Markov chains, including:

[0010] S1, prepare data and obtain detection data within round M;

[0011] S2, Based on the data alignment algorithm, perform data alignment on the detection data within the M rounds to obtain aligned detection data within multiple rounds;

[0012] S3, New Defect Identification, including: Identifying new defects relative to the previous round based on the aligned multi-round detection data, wherein the new defects cannot be determined to be active corrosion;

[0013] S4, Growth Defect Identification, including: performing growth defect analysis on the newly added defects based on the overlap coefficient algorithm and the aligned multi-round detection data, and identifying whether the growth defect of the pipeline in the current round of detection relative to the previous round is active corrosion or inactive corrosion.

[0014] S5. Based on the M rounds of pipeline inspection data, a Markov chain state transition matrix is ​​constructed to intelligently predict the defect growth during the next round of pipeline inspection.

[0015] S6 performs an intrinsic safety assessment of pipeline defects based on intelligent predictions of defect growth during the next round of pipeline inspection, and makes corresponding maintenance and repair decisions.

[0016] Preferably, S1 includes:

[0017] S11, acquire inspection data within M rounds, including acquiring inspection data from the first round and inspection data from the second to the Mth rounds (a total of M-1 rounds); and the acquisition of the total M-1 rounds of inspection data refers to the inspection data obtained from the same pipeline segment and the same inspection technology corresponding to the acquisition of all previous rounds of inspection data for this round; wherein, the first round of inspection data includes: defect location, defect type, and defect detection time, the defect type includes pipeline corrosion depth and crack length, and the same inspection technology includes MFL inspection technology or ultrasonic inspection technology;

[0018] S12, standardize the detection data within the M-round to improve the accuracy of data alignment.

[0019] Preferably, the data alignment algorithm includes a linear stretching algorithm, a KL divergence algorithm, and a semantic similarity algorithm.

[0020] Preferably, S4 includes:

[0021] S41, calculate the coincidence coefficient of the probability density function of the true size distribution of defects in multiple internal inspections. The specific formula is shown in equation (1) below:

[0022] In the formula, represents the overlap coefficient, indicating the degree of overlap between the probability density functions of the two true size distributions. The probability density function of the true size distribution of the defect detected in the first internal inspection is , and the probability density function of the true size distribution of the defect detected in the second internal inspection is . The function value is determined by the normal distribution function that returns the specified mean and standard deviation, as shown in the formula.

[0023]

[0024] Where X represents the numerical value of the true size distribution of internally detected defects; mean represents the mean of the true size distribution of internally detected defects; standard-dev represents the standard deviation of the true size distribution of internally detected defects; and cumulative represents the logical value that determines the form of the normal distribution function. If cumulative is TRUE, NORMDIST returns the cumulative distribution function; if it is FALSE, it returns the probability density function.

[0025] S42, based on the coincidence coefficient, perform a defect growth significance analysis on the newly added defects to identify the defects that have grown relative to the previous round when the pipeline is inspected in the current round, including: calculating the coincidence coefficient of the probability density function of the true size distribution of defects in the two rounds of inspection; if the coincidence coefficient is less than or equal to P, it is considered that the defect has grown and is active corrosion; if the coincidence coefficient is greater than P, it is considered that the defect has not grown and is inactive corrosion.

[0026] Preferably, S5 includes:

[0027] S51, based on the M-round inspection data of the pipeline, calculate the defect overlap coefficient corresponding to the growth defect situation of the inspection data defects in each round; calculate the growth defect situation of the inspection data defects in the first round and the second round, and denote the defect overlap coefficient as... ; Calculate the growth of defects in the detection data between the second and third rounds, and denote the defect overlap coefficient as . ;

[0028] S52, based on the defect overlap coefficient corresponding to the defect growth of the inspection data in each round, calculates the state transition frequency of each defect, and constructs a Markov chain state transition matrix based on this, so as to intelligently predict the defect growth in the next round of pipeline inspection.

[0029] Preferably, S52 includes: Based on the defect coincidence coefficient, the frequency of each defect state transition is counted as the defect state change.

[0030] (2) Based on the frequency of each defect state transition, obtain the total number of defects that grow in each round, the total number of defects that do not grow, the total number of defects that grow in all rounds, the total number of defects that do not grow in any round, the total number of defects that do not grow in some rounds but grow in other rounds, and the total number of defects that grow in some rounds but do not grow in other rounds; use the above total number of defects as the data basis for constructing the Markov chain state transition matrix; denote the total number of defects that grow in the first and second rounds as Y, the total number of defects that do not grow in the first and second rounds as N, the total number of defects that grow in the first and second rounds and also grow in the second and third rounds as Y1, the defects that grow in the first and second rounds but do not grow in the second and third rounds as N1; denote the defects that do not grow in the first and second rounds but grow in the second and third rounds as Y2, and the defects that do not grow in the first and second rounds and also do not grow in the second and third rounds as N2;

[0031] (3) Construct the Markov chain state transition matrix. The Markov chain state transition matrix is ​​constructed using the statistical data in Table 1. The construction method is shown in Equation (3):

[0032]

[0033] Among them, the total number of defects that increased in the first and second rounds is denoted as Y, the total number of defects that did not increase in the first and second rounds is denoted as N, the total number of defects that increased in the first and second rounds and also increased in the second and third rounds is denoted as Y1, and the defects that increased in the first and second rounds but did not increase in the second and third rounds are denoted as N1; the defects that did not increase in the first and second rounds but increased in the second and third rounds are denoted as Y2, and the defects that did not increase in the first and second rounds and did not increase in the second and third rounds are denoted as N2; p 11 p is the defect overlap coefficient between Y and Y1. 12 p is the defect overlap coefficient between Y and N1. 21 p is the defect overlap coefficient between Y and Y2. 11 The defect overlap coefficient between N and N2;

[0034] (4) Construct the initial matrix corresponding to the Markov chain state transition matrix to predict and analyze the growth of defects in the fourth round. The construction of the initial matrix includes: determining the defect change state in the current round of the predicted defect compared to the previous round. If a defect state in the second or third round is increasing, then the initial matrix is... When a defect state in the second or third round is not growing, then the initial matrix... ;

[0035] (5) Using the detection data in the (M+1)th round as the validation set, the growth of pipeline defects in the same pipeline segment is calibrated and predicted, wherein the calibration prediction is used for model calibration; according to the Markov chain theory, the state V1 of a certain defect in the fourth round of detection is When the initial matrix of a predicted defect is If P is in the calculation result of V1 V1 >P V2If the defect status is predicted to increase within the third or fourth round of detection, then the defect status is predicted to not increase; otherwise, it is predicted to not increase. When the initial matrix of a certain defect is predicted... If P is in the calculation result of V1 V1 >P V2 If the defect status is not increased in the third or fourth round of inspection, then the defect status is predicted to be not increased; otherwise, it is predicted to be increased.

[0036] Preferably, S6 includes:

[0037] S61, determine the inherent safety risk assessment indicators for pipeline defects, including: defect growth probability, toxicity of the medium inside the pipeline, population density around the pipeline, and maintenance difficulty;

[0038] S62, construct a corresponding pipeline defect inherent safety assessment model based on the pipeline defect inherent safety risk assessment index; wherein, the calculation formula of the pipeline defect inherent safety assessment model is:

[0039] Risk = α1 × Defect growth probability + α2 × Pipeline medium toxicity + α3 × Population density around the pipeline + α4 × Maintenance and repair difficulty; where α1, α2, α3, and α4 are the weighting coefficients for defect growth probability, pipeline medium toxicity, population density around the pipeline, and maintenance and repair difficulty.

[0040] S63, The weight coefficients α1, α2, α3 and α4 of the pipeline intrinsic safety risk assessment index are determined by the analytic hierarchy process (AHP);

[0041] S63 includes:

[0042] (1) The weights of each dimension are determined by the Analytic Hierarchy Process (AHP), and a judgment matrix is ​​constructed;

[0043] (2) The weights are calculated using the "sum-product method", and the specific steps are as follows:

[0044] A. Based on the normalization principle, the column is normalized as shown in equation (4):

[0045] ;

[0046] B. Add the values ​​of each row after column normalization to obtain a sum vector, which is a column vector, as shown in equation (5):

[0047]

[0048] Income That is, the weight of each influencing factor;

[0049] (3) Calculate the largest eigenvalue of the judgment matrix;

[0050] Firstly, calculate This results in a column vector, then... and The sum of the corresponding values ​​divided by m yields the largest eigenvalue, as shown in equation (6):

[0051]

[0052] (4) Perform a consistency check on the judgment matrix:

[0053] The formula for calculating the consistency index in the consistency test is shown in equation (7):

[0054]

[0055] In the formula: This is a consistency index, representing the relative error with respect to m;

[0056] The formula for calculating the consistency ratio is shown in equation (8):

[0057] (8)

[0058] In the formula: Consistency ratio; As an average consistency indicator, if the consistency ratio If the ratio is less than 0.1, the relative importance comparison matrix is ​​considered to have good consistency. If the value is greater than 0.1, the judgment matrix with relatively high importance needs to be readjusted and the calculations in equations (7) and (8) are performed again until satisfactory consistency is achieved. Then, the weight vector W=[α1,α2,α3,α4] is obtained through the eigenvector calculation, and the weight coefficients α1, α2, α3 and α4 of the pipeline intrinsic safety risk assessment index are determined.

[0059] S64. Risk assessment indicators are scored, including four indicators: probability of defect growth, toxicity of pipeline medium, population density around pipeline, and difficulty of maintenance and repair. Each indicator is scored out of 100, and the scores for probability of defect growth, toxicity of pipeline medium, population density around pipeline, and difficulty of maintenance and repair are A1, A2, A3, and A4.

[0060] S65, obtain the final single defect risk score: Risk=α1×A1+α2×A2+α3×A3+α4×A4;

[0061] S66. Based on the final single-defect risk score, determine whether the pipeline defect is high-risk, medium-risk, or low-risk, and determine the corresponding maintenance and repair decision.

[0062] A second aspect of the present invention is to provide a pipeline defect growth prediction system based on multi-round intra-inspection data for implementing the method of the first aspect, comprising:

[0063] The data acquisition module (101) is used to prepare data and acquire detection data within round M.

[0064] The data alignment module (102) is used to perform data alignment on the detection data within the M rounds based on the data alignment algorithm to obtain the aligned detection data within the multiple rounds.

[0065] New defect identification (103) is used to identify new defects relative to the previous round based on the aligned multi-round detection data, and it is impossible to determine whether the new defects are active corrosion;

[0066] The growth defect identification module (104) is used to perform growth defect analysis on the newly added defect based on the overlap coefficient algorithm and the aligned multi-round detection data, and to identify whether the growth defect of the pipeline in the current round of detection is active corrosion or inactive corrosion relative to the previous round.

[0067] The intelligent prediction module for defect growth (105) is used to construct a Markov chain state transition matrix based on the M rounds of pipeline inspection data, so as to intelligently predict the defect growth in the next round of pipeline inspection.

[0068] The pipeline defect inherent safety assessment module (106) is used to conduct pipeline defect inherent safety assessment based on intelligent prediction of defect growth during the next round of pipeline inspection, and to make corresponding maintenance and repair decisions.

[0069] A third aspect of the present invention is to provide an electronic device including a processor and a memory, the memory storing a plurality of instructions, the processor being configured to read the instructions and execute the method as described in the second aspect.

[0070] A fourth aspect of the present invention is to provide a computer-readable storage medium storing a plurality of instructions which can be read by a processor and executed as described in the second aspect.

[0071] The beneficial effects of the method and system of the present invention are as follows:

[0072] This invention provides a method and system for predicting pipeline defect growth based on multi-round inspection data. Belonging to the category of statistical defect growth prediction methods, this invention statistically analyzes the defect growth status of pipeline inspection data across three rounds, constructs a defect state transition matrix, quantifies the randomness of defect evolution, and predicts the growth status of defects in the next round. Defects predicted to be growing are given priority attention, while those predicted not to be growing are given less attention. This achieves high-precision predictive maintenance decision support, applicable to complex scenarios such as long-distance pipelines and subsea pipelines. The proposed method can significantly reduce the economic burden on enterprises in pipeline intrinsic safety analysis and provide data support for pipeline operation and maintenance management.

[0073] The above and other objects, advantages and features of the present invention will become more apparent to those skilled in the art from the following detailed description of specific embodiments of the invention in conjunction with the accompanying drawings. Attached Figure Description

[0075] The following description will detail some specific embodiments of the invention by way of example and not limitation, with reference to the accompanying drawings. The same reference numerals in the drawings denote the same or similar parts or portions. Those skilled in the art will understand that these drawings are not necessarily drawn to scale. The objectives and features of the invention will become more apparent from the following description taken in conjunction with the accompanying drawings, in which:

[0076] Figure 1 This is a schematic diagram of the pipeline defect growth prediction method based on multi-round internal inspection data according to an embodiment of the present invention;

[0077] Figure 2 This is a schematic diagram of the growth defect identification process according to an embodiment of the present invention;

[0078] Figure 3 This is a schematic diagram of the defect growth prediction method based on Markov chains according to an embodiment of the present invention.

[0079] Figure 4 This is an architecture diagram of a pipeline defect growth prediction system based on multi-round internal detection data according to an embodiment of the present invention.

[0080] Figure 5 This is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0082] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0083] Example 1

[0084] like Figure 1 As shown, this embodiment provides a pipeline defect growth prediction method based on multi-round intra-inspection data. It dynamically predicts pipeline defect growth based on multi-round intra-inspection data (ILI) and a Markov chain, including:

[0085] S1, prepare data and obtain detection data within round M;

[0086] In a preferred embodiment, S1 includes:

[0087] S11, acquire inspection data within M rounds, including acquiring the first round of inspection data and the inspection data of M-1 rounds from the second to the Mth round; and the acquisition of the total M-1 rounds of inspection data is the same pipeline segment and the same inspection technology obtained corresponding to the acquisition of all previous rounds of inspection data for this round; wherein, the first round of inspection data includes: defect location, defect type and defect detection time, the defect type includes pipeline corrosion depth and crack length, and the same inspection technology includes MFL inspection technology or ultrasonic inspection technology.

[0088] As a preferred implementation, the Markov chain-based defect growth prediction method requires at least three rounds of pipeline inspection data. Based on this data, the growth of pipeline defects during the fourth round of inspection is predicted. Specifically, as follows:

[0089] • First round of inspection data (P1): including defect location, type (e.g., corrosion depth, crack length), and inspection time;

[0090] • Second round of inspection data (P2): Inspection results of the same pipeline section and the same inspection technology (such as MFL, ultrasonic) as P1;

[0091] • Third round of testing data (P3): Testing results for the same pipe section and using the same testing technology (such as MFL, ultrasonic) as P1 and P2;

[0092] • Fourth round of testing data (P4): Used as a validation set for model calibration.

[0093] S12, standardize the detection data within the M-round to improve the accuracy of data alignment;

[0094] In this embodiment, after the data is prepared, the format of each internal detection result needs to be standardized to ensure the accuracy of subsequent data alignment and application.

[0095] S2, Based on the data alignment algorithm, perform data alignment on the detection data within the M rounds to obtain aligned detection data within multiple rounds;

[0096] As a preferred embodiment, the data alignment algorithm includes a linear stretching algorithm, a KL divergence algorithm, and a semantic similarity algorithm.

[0097] In this embodiment, linear stretching algorithm, KL divergence algorithm, semantic similarity algorithm and other methods are used to align the acquired pipeline three-round detection data.

[0098] S3, New Defect Identification, including: Identifying new defects relative to the previous round based on the aligned multi-round detection data, wherein the new defects cannot be determined to be active corrosion.

[0099] In this embodiment, newly added defects relative to the previous round are identified based on the aligned data. These defects indicate that a new defect has appeared in the pipeline at the newly added defect location, but it cannot be determined whether they are active defects. Therefore, according to the data acquisition principle, when a new defect appears at a certain point in the pipeline, it is necessary to wait until the defect at that point has been detected for three rounds before the proposed new defect prediction method based on Markov chains can be used for analysis.

[0100] S4, Growth Defect Identification, includes: performing growth defect analysis on the newly added defects based on the overlap coefficient algorithm and the aligned multi-round detection data, identifying whether the growth defect in the pipeline during the current round of detection is active corrosion or inactive corrosion relative to the previous round. The growth defect identification uses the overlap coefficient algorithm, and the application process is as follows: Figure 2 As shown.

[0101] As a preferred implementation, S4 includes:

[0102] S41, calculate the coincidence coefficient of the probability density function of the true size distribution of defects in multiple internal inspections. The specific formula is shown in equation (1) below:

[0103]

[0104] In the formula: The overlap coefficient represents the degree of overlap between the probability density functions of two true size distributions. Let be the probability density function of the true size distribution of the defects detected in the first internal inspection. Let be the probability density function of the true size distribution of the defects detected in the second internal inspection. The function value is determined by the normal distribution function that returns the specified mean and standard deviation, as shown in equation (2):

[0105] ;

[0106] Where x represents the numerical value of the true size distribution of internally detected defects; mean represents the mean of the true size distribution of internally detected defects; standard-dev represents the standard deviation of the true size distribution of internally detected defects; and cumulative represents the logical value that determines the form of the normal distribution function. If cumulative is TRUE, NORMDIST returns the cumulative distribution function; if it is FALSE, it returns the probability density function.

[0107] S42, based on the coincidence coefficient, perform a defect growth significance analysis on the newly added defects to identify the defects that have grown relative to the previous round when the pipeline is inspected in the current round, including: calculating the coincidence coefficient of the probability density function of the true size distribution of defects in the two rounds of inspection; if the coincidence coefficient is less than or equal to P, it is considered that the defect has grown and is active corrosion; if the coincidence coefficient is greater than P, it is considered that the defect has not grown and is inactive corrosion.

[0108] As a preferred embodiment, P is set to 0.5. Of course, those skilled in the art should know that it can also be set to other appropriate values ​​as needed, all of which are within the protection scope of this invention.

[0109] S5. Based on the M rounds of pipeline inspection data, a Markov chain state transition matrix is ​​constructed to intelligently predict the defect growth during the next round of pipeline inspection.

[0110] In a preferred embodiment, S5 includes:

[0111] S51, Calculate the defect overlap coefficient corresponding to the growth defect situation of defects in each round based on the M round inspection data of the pipeline;

[0112] In this embodiment, the growth of defects in the detected data during the first and second rounds is calculated, and the defect overlap coefficient is denoted as... ; Calculate the growth of defects in the detection data between the second and third rounds, and denote the defect overlap coefficient as . .

[0113] like Figure 3 As shown in S52, based on the defect overlap coefficient corresponding to the defect growth of the inspection data in each round, the frequency of each defect state transition is counted, and a Markov chain state transition matrix is ​​constructed on this basis, so as to intelligently predict the defect growth in the next round of pipeline inspection.

[0114] In a preferred embodiment, S52 includes:

[0115] (1) Based on the defect overlap coefficient, the frequency of each defect state transition is counted as the defect state change;

[0116] (2) Based on the state transition frequency of each defect, obtain the total number of defects that grow in each round, the total number of defects that do not grow, the total number of defects that grow in each round, the total number of defects that do not grow in each round, the total number of defects that do not grow in some rounds but grow in other rounds, and the total number of defects that grow in some rounds but do not grow in other rounds; use the above total number of defects as the data basis for constructing the Markov chain state transition matrix.

[0117] In this embodiment, the total number of defects that increase in the first and second rounds is denoted as Y, the total number of defects that do not increase in the first and second rounds is denoted as N, the total number of defects that increase in the first and second rounds and also increase in the second and third rounds is denoted as Y1, the defects that increase in the first and second rounds but do not increase in the second and third rounds is denoted as N1; the defects that do not increase in the first and second rounds but increase in the second and third rounds are denoted as Y2, and the defects that do not increase in the first and second rounds and do not increase in the second and third rounds are denoted as N2.

[0118] In this embodiment, according to the above data recording criteria, the data after calculating the newly added defects in the pipeline inspection data within three rounds is recorded. The calculation results can be recorded according to Table 1, the record table of increased / unincreased defects in the inspection data within three rounds.

[0119] Table 1

[0120]

[0121] (3) Construct the Markov chain state transition matrix. The Markov chain state transition matrix is ​​constructed using the statistical data in Table 1. The construction method is shown in Equation (3):

[0122]

[0123] Among them, the total number of defects that increased in the first and second rounds is denoted as Y, the total number of defects that did not increase in the first and second rounds is denoted as N, the total number of defects that increased in the first and second rounds and also increased in the second and third rounds is denoted as Y1, and the defects that increased in the first and second rounds but did not increase in the second and third rounds are denoted as N1; the defects that did not increase in the first and second rounds but increased in the second and third rounds are denoted as Y2, and the defects that did not increase in the first and second rounds and did not increase in the second and third rounds are denoted as N2; p 11 p is the defect overlap coefficient between Y and Y1. 12 p is the defect overlap coefficient between Y and N1. 21 p is the defect overlap coefficient between Y and Y2. 11 The defect overlap coefficient is between N and N2.

[0124] (4) Construct the initial matrix corresponding to the Markov chain state transition matrix to predict and analyze the growth of defects in the fourth round. The construction of the initial matrix includes: determining the defect change state in the current round of the predicted defect compared to the previous round. If a defect state in the second or third round is increasing, then the initial matrix is... When a defect state in the second or third round is not growing, then the initial matrix... ;

[0125] (5) Using the detection data in the (M+1)th round as the validation set, the growth of pipeline defects in the same pipeline segment is calibrated and predicted, wherein the calibration prediction is used for model calibration;

[0126] In this embodiment, according to Markov chain theory, the state V1 of a certain defect during the fourth round of detection is: When the initial matrix of a certain defect to be predicted is If P is in the calculation result of V1 V1 >P V2 If the defect status is predicted to increase within the third or fourth round of detection, then the defect status is predicted to not increase; otherwise, it is predicted to not increase. When the initial matrix of a certain defect is predicted... If P is in the calculation result of V1 V1 >P V2 If the defect status is not increased in the third or fourth round of inspection, then the defect status is predicted to be not increased; otherwise, it is predicted to be increased.

[0127] In a preferred embodiment, the method further includes:

[0128] S6 performs an intrinsic safety assessment of pipeline defects based on intelligent predictions of defect growth during the next round of pipeline inspection, and makes corresponding maintenance and repair decisions.

[0129] In a preferred embodiment, S6 includes:

[0130] S61 defines the inherent safety risk assessment indicators for pipeline defects, including: defect growth probability, toxicity of the medium inside the pipeline, population density around the pipeline, and maintenance difficulty. Defect growth probability considers that when the growth probability of a defect is greater than 0.5, it indicates a higher probability that the defect is active and will pose a greater risk in the future. Toxicity of the medium inside the pipeline considers that if a corrosion leak occurs due to the defect, and the medium is highly toxic, it will have a serious impact on surrounding residents and the environment. Population density considers that leaks in densely populated areas pose a greater risk of casualties and economic losses than in other areas. Maintenance difficulty considers that the difficulty of maintaining and repairing defects varies depending on their location, surrounding environment, and operating conditions after a leak.

[0131] S62, construct a corresponding pipeline defect inherent safety assessment model based on the pipeline defect inherent safety risk assessment index; wherein, the calculation formula of the pipeline defect inherent safety assessment model is:

[0132] Risk = α1 × Defect growth probability + α2 × Pipeline medium toxicity + α3 × Population density around the pipeline + α4 × Maintenance and repair difficulty; where α1, α2, α3, and α4 are the weighting coefficients for defect growth probability, pipeline medium toxicity, population density around the pipeline, and maintenance and repair difficulty.

[0133] S63, The weight coefficients α1, α2, α3 and α4 of the pipeline intrinsic safety risk assessment index are determined by the analytic hierarchy process (AHP);

[0134] In this embodiment, S63 includes:

[0135] (1) The weights of each dimension are determined using the Analytic Hierarchy Process (AHP), and a judgment matrix is ​​constructed, as shown in Table 2: Analytic Hierarchy Process Judgment Matrix

[0136] Table 2

[0137]

[0138] (2) The weights are calculated using the "sum-product method", and the specific steps are as follows:

[0139] A. Based on the normalization principle, the columns of Table 2 are normalized as shown in equation (4):

[0140]

[0141] B. Add the values ​​of each row after column normalization to obtain a sum vector, which is a column vector, as shown in equation (5):

[0142]

[0143] Income This refers to the weight of each influencing factor.

[0144] (3) Calculate the largest eigenvalue of the judgment matrix;

[0145] First calculate This results in a column vector, then... and The sum of the corresponding values ​​divided by m yields the largest eigenvalue, as shown in equation (6):

[0146]

[0147] (4) Perform a consistency check on the judgment matrix:

[0148] The consistency test is to verify the reliability of the judgment matrix. The judgment matrix is ​​constructed by the Analytic Hierarchy Process (AHP), which essentially uses a scoring method, and therefore has a certain degree of subjectivity, which may result in some error compared with the theoretical matrix. The consistency test examines the degree of this error, thereby determining whether the established judgment matrix has satisfactory consistency; the formula for calculating the consistency index of the consistency test is shown in equation (7):

[0149]

[0150] In the formula: This is a consistency index, representing the relative error with respect to m;

[0151] The formula for calculating the consistency ratio is shown in equation (8):

[0152] (8)

[0153] In the formula: Consistency ratio; The average consistency index can be found in Table 3, Average Random Consistency Index. If the consistency ratio... If the ratio is less than 0.1, the relative importance comparison matrix is ​​considered to have good consistency. If the value is greater than 0.1, the judgment matrix with relatively high importance needs to be readjusted and the calculations in equations (7) and (8) are performed again until satisfactory consistency is achieved. Then, the weight vector W=[α1,α2,α3,α4] is obtained through the eigenvector calculation, and the weight coefficients α1, α2, α3 and α4 of the pipeline intrinsic safety risk assessment index are determined.

[0154] Table 3

[0155]

[0156] S64. Risk assessment indicators are scored, including four indicators: probability of defect growth, toxicity of pipeline medium, population density around pipeline, and difficulty of maintenance and repair. Each indicator is scored out of 100, and the scores for probability of defect growth, toxicity of pipeline medium, population density around pipeline, and difficulty of maintenance and repair are A1, A2, A3, and A4.

[0157] In this embodiment, the specific scoring method is illustrated in Table 4, the risk assessment indicator scoring table:

[0158] Table 4

[0159]

[0160] S65, obtain the final single defect risk score: Risk=α1×A1+α2×A2+α3×A3+α4×A4;

[0161] In this embodiment, the inherent safety risk assessment results of the defects are shown in Table 5.

[0162] Table 5

[0163]

[0164] S66. Based on the final single-defect risk score, determine whether the pipeline defect is high-risk, medium-risk, or low-risk, and determine the corresponding maintenance and repair decision.

[0165] In this embodiment, based on the inherent safety evaluation results of pipeline defects in Table 5, targeted maintenance and repair decisions are made, as detailed below:

[0166] High risk: Immediately stop and repair the service;

[0167] Medium risk: Maintenance is planned based on actual production conditions;

[0168] • Low risk: Continuous monitoring, but not a priority focus.

[0169] Example 2

[0170] like Figure 4 As shown, this embodiment provides a pipeline defect growth prediction system based on multi-round intra-inspection data, used to implement the method of Embodiment 1, including:

[0171] Data acquisition module 101 is used for data preparation and to acquire detection data within round M.

[0172] Data alignment module 102 is used to perform data alignment on the detection data within the M rounds based on a data alignment algorithm to obtain aligned detection data within multiple rounds;

[0173] The newly added defect identification 103 is used to identify new defects relative to the previous round based on the aligned multi-round detection data, and it is impossible to determine whether the new defects are active corrosion.

[0174] The growth defect identification module 104 is used to perform growth defect analysis on the newly added defect based on the overlap coefficient algorithm and the aligned multi-round detection data, and to identify whether the growth defect of the pipeline in the current round of detection is active corrosion or inactive corrosion relative to the previous round.

[0175] The intelligent prediction module 105 for defect growth is used to construct a Markov chain state transition matrix based on the pipeline's M-round inspection data, thereby intelligently predicting the defect growth during the next round of pipeline inspection.

[0176] The pipeline defect inherent safety assessment module 106 is used to conduct an inherent safety assessment of pipeline defects based on intelligent prediction of defect growth during the next round of pipeline inspection, and to make corresponding maintenance and repair decisions.

[0177] The present invention also provides a memory that stores multiple instructions for implementing the method as described in Embodiment 1.

[0178] like Figure 5As shown, the present invention also provides an electronic device, including a processor 301 and a memory 302 connected to the processor 301. The memory 302 stores a plurality of instructions, which can be loaded and executed by the processor to enable the processor to perform the method as described in Embodiment 1.

[0179] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A method for predicting pipeline defect growth based on multi-round internal inspection data, characterized in that, Dynamic prediction of pipeline defect growth based on multi-round internal inspection data and Markov chains, including: S1, prepare data and obtain detection data within round M; S2, Based on the data alignment algorithm, perform data alignment on the detection data within the M rounds to obtain aligned detection data within multiple rounds; S3, New Defect Identification, including: Identifying new defects relative to the previous round based on the aligned multi-round detection data, wherein the new defects cannot be determined to be active corrosion; S4, Growth Defect Identification, including: performing growth defect analysis on the newly added defects based on the overlap coefficient algorithm and the aligned multi-round detection data, and identifying whether the growth defect of the pipeline in the current round of detection relative to the previous round is active corrosion or inactive corrosion. S5. Based on the M rounds of pipeline inspection data, a Markov chain state transition matrix is ​​constructed to intelligently predict the defect growth during the next round of pipeline inspection. S6 performs an intrinsic safety assessment of pipeline defects based on intelligent predictions of defect growth during the next round of pipeline inspection, and makes corresponding maintenance and repair decisions.

2. The pipeline defect growth prediction method based on multi-round intra-inspection data according to claim 1, characterized in that, S1 includes: S11, acquire inspection data within M rounds, including acquiring inspection data from the first round and inspection data from the second to the Mth rounds (a total of M-1 rounds); and the acquisition of the total M-1 rounds of inspection data refers to the inspection data obtained from the same pipeline segment and the same inspection technology corresponding to the acquisition of all previous rounds of inspection data for this round; wherein, the first round of inspection data includes: defect location, defect type, and defect detection time, the defect type includes pipeline corrosion depth and crack length, and the same inspection technology includes MFL inspection technology or ultrasonic inspection technology; S12, standardize the detection data within the M-round to improve the accuracy of data alignment.

3. The pipeline defect growth prediction method based on multi-round intra-inspection data according to claim 2, characterized in that, The data alignment algorithms include linear stretching algorithm, KL divergence algorithm and semantic similarity algorithm.

4. The pipeline defect growth prediction method based on multi-round intra-inspection data according to claim 3, characterized in that, S4 includes: S41, calculate the coincidence coefficient of the probability density function of the true size distribution of defects in multiple internal inspections. The specific formula is shown in equation (1) below: ; In the formula: The overlap coefficient represents the degree of overlap between the probability density functions of two true size distributions. Let be the probability density function of the true size distribution of the defects detected in the first internal inspection. Let be the probability density function of the true size distribution of the defects detected in the second internal inspection. , The function value is determined by the normal distribution function that returns the specified mean and standard deviation, as shown in equation (2): ; Where X represents the numerical value of the true size distribution of internally detected defects; mean represents the mean of the true size distribution of internally detected defects; standard-dev represents the standard deviation of the true size distribution of internally detected defects; and cumulative represents the logical value that determines the form of the normal distribution function. If cumulative is TRUE, NORMDIST returns the cumulative distribution function; if it is FALSE, it returns the probability density function. S42, based on the coincidence coefficient, perform a defect growth significance analysis on the newly added defects to identify the defects that have grown relative to the previous round when the pipeline is inspected in the current round, including: calculating the coincidence coefficient of the probability density function of the true size distribution of defects in the two rounds of inspection; if the coincidence coefficient is less than or equal to P, it is considered that the defect has grown and is active corrosion; if the coincidence coefficient is greater than P, it is considered that the defect has not grown and is inactive corrosion.

5. The pipeline defect growth prediction method based on multi-round intra-inspection data according to claim 4, characterized in that, S5 includes: S51, based on the M-round inspection data of the pipeline, calculate the defect overlap coefficient corresponding to the growth defect situation of the inspection data defects in each round; calculate the growth defect situation of the inspection data defects in the first round and the second round, and denote the defect overlap coefficient as... ; Calculate the growth of defects in the detection data between the second and third rounds, and denote the defect overlap coefficient as . ; S52, based on the defect overlap coefficient corresponding to the defect growth of the inspection data in each round, calculates the state transition frequency of each defect, and constructs a Markov chain state transition matrix based on this, so as to intelligently predict the defect growth in the next round of pipeline inspection.

6. The pipeline defect growth prediction method based on multi-round intra-inspection data according to claim 5, characterized in that, S52 includes: (1) Based on the defect coincidence coefficient, the frequency of each defect state transition is counted as the defect state change; (2) Based on the state transition frequencies of each defect, obtain the total number of defects that grow in each round, the total number of defects that do not grow, the total number of defects that grow in all rounds, the total number of defects that do not grow in any round, the total number of defects that do not grow in some rounds but grow in other rounds, and the total number of defects that grow in some rounds but do not grow in other rounds; use the above total number of defects as the data basis for constructing the Markov chain state transition matrix; denote the total number of defects that grow in the first and second rounds as Y, the total number of defects that do not grow in the first and second rounds as N, the total number of defects that grow in the first and second rounds and also grow in the second and third rounds as Y1, the defects that grow in the first and second rounds but do not grow in the second and third rounds as N1; denote the defects that do not grow in the first and second rounds but grow in the second and third rounds as Y2, and the defects that do not grow in the first and second rounds and also do not grow in the second and third rounds as N2; (3) Construct the Markov chain state transition matrix. The Markov chain state transition matrix is ​​constructed using the statistical data in Table 1. The construction method is shown in Equation (3): Among them, the total number of defects that increased in the first and second rounds is denoted as Y, the total number of defects that did not increase in the first and second rounds is denoted as N, the total number of defects that increased in the first and second rounds and also increased in the second and third rounds is denoted as Y1, and the defects that increased in the first and second rounds but did not increase in the second and third rounds are denoted as N1; the defects that did not increase in the first and second rounds but increased in the second and third rounds are denoted as Y2, and the defects that did not increase in the first and second rounds and did not increase in the second and third rounds are denoted as N2; p 11 p is the defect overlap coefficient between Y and Y1. 12 p is the defect overlap coefficient between Y and N1. 21 p is the defect overlap coefficient between Y and Y2. 11 The defect overlap coefficient between N and N2; (4) Construct the initial matrix corresponding to the Markov chain state transition matrix to predict and analyze the growth of defects in the fourth round. The construction of the initial matrix includes: determining it based on the defect change state of the current round compared to the previous round. If a defect state in the second or third round is increasing, then the initial matrix is... When a defect state in the second or third round is not growing, then the initial matrix... (5) Using the detection data in the (M+1)th round as the validation set, the growth of pipeline defects in the same pipeline segment is calibrated and predicted, wherein the calibration prediction is used for model calibration; according to the Markov chain theory, the state V1 of a certain defect in the fourth round of detection is... When the initial matrix of a certain defect to be predicted is If P is in the calculation result of V1 V1 >P V2 If the defect status is predicted to increase within the third or fourth round of detection, then the defect status is predicted to not increase; otherwise, it is predicted to not increase. When the initial matrix of a certain defect is predicted... If P is in the calculation result of V1 V1 >P V2 If the defect status is not increased in the third or fourth round of inspection, then the defect status is predicted to be not increased; otherwise, it is predicted to be increased.

7. The pipeline defect growth prediction method based on multi-round intra-inspection data according to claim 6, characterized in that, S6 includes: S61, determine the inherent safety risk assessment indicators for pipeline defects, including: defect growth probability, toxicity of the medium inside the pipeline, population density around the pipeline, and maintenance difficulty; S62, construct a corresponding pipeline defect inherent safety assessment model based on the pipeline defect inherent safety risk assessment index; wherein, the calculation formula of the pipeline defect inherent safety assessment model is: Risk = α1 × Defect growth probability + α2 × Pipeline medium toxicity + α3 × Population density around the pipeline + α4 × Maintenance and repair difficulty; where α1, α2, α3, and α4 are the weighting coefficients for defect growth probability, pipeline medium toxicity, population density around the pipeline, and maintenance and repair difficulty. S63, The weight coefficients α1, α2, α3 and α4 of the pipeline intrinsic safety risk assessment index are determined by the analytic hierarchy process (AHP); S63 includes: (1) The weights of each dimension are determined by the Analytic Hierarchy Process (AHP), and a judgment matrix is ​​constructed; (2) The weights are calculated using the "sum-product method", and the specific steps are as follows: A. Based on the normalization principle, the column is normalized as shown in equation (4): ； B. Add the values ​​of each row after column normalization to obtain a sum vector, which is a column vector, as shown in equation (5): ; income That is, the weight of each influencing factor; (3) Calculate the largest eigenvalue of the judgment matrix; Firstly, calculate This results in a column vector, then... and The sum of the corresponding values ​​divided by m yields the largest eigenvalue, as shown in equation (6): ； (4) Perform a consistency check on the judgment matrix: The formula for calculating the consistency index in the consistency test is shown in equation (7). ; In the formula: As a consistency indicator, it represents the agreement between the two sides. The relative error; The formula for calculating the consistency ratio is shown in equation (8): （8）； In the formula: The consistency ratio; As an average consistency indicator, if the consistency ratio If the ratio is less than 0.1, the relative importance comparison matrix is ​​considered to have good consistency. If the value is greater than 0.1, the judgment matrix with relatively high importance needs to be readjusted and the calculations in equations (7) and (8) are performed again until satisfactory consistency is achieved. Then, the weight vector W=[α1,α2,α3,α4] is obtained through the eigenvector calculation, and the weight coefficients α1, α2, α3 and α4 of the pipeline intrinsic safety risk assessment index are determined. S64. Risk assessment indicators are scored, including four indicators: probability of defect growth, toxicity of pipeline medium, population density around pipeline, and difficulty of maintenance and repair. Each indicator is scored out of 100, and the scores for probability of defect growth, toxicity of pipeline medium, population density around pipeline, and difficulty of maintenance and repair are A1, A2, A3, and A4. S65, obtain the final single defect risk score: Risk=α1×A1+α2×A2+α3×A3+α4×A4; S66. Based on the final single-defect risk score, determine whether the pipeline defect is high-risk, medium-risk, or low-risk, and determine the corresponding maintenance and repair decision.

8. A pipeline defect growth prediction system based on multi-round intra-inspection data, used to implement the method according to any one of claims 1-7, characterized in that, include: The data acquisition module (101) is used to prepare data and acquire detection data within round M. The data alignment module (102) is used to perform data alignment on the detection data within the M rounds based on the data alignment algorithm to obtain the aligned detection data within the multiple rounds. New defect identification (103) is used to identify new defects relative to the previous round based on the aligned multi-round detection data, and it is impossible to determine whether the new defects are active corrosion; The growth defect identification module (104) is used to perform growth defect analysis on the newly added defect based on the overlap coefficient algorithm and the aligned multi-round detection data, and to identify whether the growth defect of the pipeline in the current round of detection is active corrosion or inactive corrosion relative to the previous round. The intelligent prediction module for defect growth (105) is used to construct a Markov chain state transition matrix based on the M rounds of pipeline inspection data, so as to intelligently predict the defect growth in the next round of pipeline inspection. The pipeline defect inherent safety assessment module (106) is used to conduct pipeline defect inherent safety assessment based on intelligent prediction of defect growth during the next round of pipeline inspection, and to make corresponding maintenance and repair decisions.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing multiple instructions, and the processor being used to read the instructions and execute the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions, which can be read by a processor and executed as described in any one of claims 1-7.

Images