A preferred method for identifying anomaly data in corrosion rates within natural gas pipelines.
By optimizing a suitable outlier identification algorithm using the analytic hierarchy process, the difficulties in identification caused by differences in sample size, data dimensionality, and normality were resolved. This enabled accurate identification of corrosion rate data within natural gas pipelines, ensuring the reliability of the analysis results.
Patent Information
- Application Number
- CN202310414290.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-04-18
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-04-18
AI Technical Summary
Existing anomaly identification algorithms differ in terms of sample size, data dimensionality, and normality, which makes it impossible to accurately identify anomalies in corrosion rates within natural gas pipelines, thus affecting the reliability of corrosion analysis results.
The analytic hierarchy process (AHP) was used to determine sample size, data dimensionality, and normality as algorithm selection criteria. By assigning weights and assigning grade scores, the most suitable algorithm for identifying abnormal data was selected. Combined with preprocessing and data organization, a database of corrosion rates in pipelines was established for identifying abnormal data.
The optimized algorithm can accurately identify abnormal data on corrosion rates inside pipelines, ensuring the authenticity of the data and laying the foundation for the reliability of corrosion analysis results.
Smart Images

Figure CN118820987B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of anomaly data identification algorithm technology, and more specifically, to a preferred method for anomaly data identification algorithm for corrosion rate inside natural gas pipelines. Background Technology
[0002] Pipelines are the primary mode of transportation for oil and natural gas. As the nation's energy arteries, oil and gas pipelines powerfully safeguard the development of the national economy. With the establishment of the National Pipeline Company and the rapid construction and commissioning of oil and gas pipelines, their role in economic development and people's lives will become even more prominent.
[0003] In the oil and gas industry, the use of sensors to record production data has led to a massive and intensive volume of data. Oil and gas pipelines generate a significant amount of data during design, construction, operation, and maintenance, providing crucial data support for analyzing pipeline condition and assessing internal corrosion. Internal corrosion rate data can be obtained through methods such as placing probes, using probes, and ultrasonic testing. With the surge in internal corrosion rate data, anomalies have become a significant problem. Subjectively, issues such as flawed sampling techniques, negligence, misrecording, or deliberate falsification can all lead to anomalies, directly impacting the quality of statistical data. Objectively, sensor malfunctions or damage can result in values significantly exceeding the average, also producing anomalies. The presence of abnormal internal corrosion rate data can cause errors in corresponding statistical analyses, significantly affecting the analysis results of pipeline corrosion. Therefore, to ensure the reliability of internal corrosion rate data, anomaly identification of the collected corrosion data is necessary.
[0004] Anomaly identification has always been a hot research topic for scholars both domestically and internationally, and it is widely used in various fields. Data mining-based anomaly identification algorithms have emerged to address this need. These methods enable automatic identification of anomaly data, improving identification efficiency and saving economic costs. Currently, commonly used anomaly identification algorithms include the Laida criterion, box plots, DBSCAN clustering, Grubbs' criterion, and KNN anomaly detection. For example, the Grubbs' criterion algorithm's core idea is to determine whether the absolute value of the residual error of a data point is greater than a corresponding critical value. If it is, the data is considered anomaly, and vice versa. This algorithm is limited to datasets with an approximately normal distribution and a sample size of less than 100. Another example is the KNN anomaly detection algorithm, which, for unknown data of a given type, identifies the K closest known data points based on distance, and then determines whether these K data points are anomalies based on their characteristics. This algorithm is suitable for large sample sizes and requires the existence of a defined normal dataset.
[0005] Therefore, different anomaly identification algorithms have varying applicability to different sample data, and the appropriate anomaly identification algorithms also differ depending on the type of corrosion data. Consequently, hastily applying an anomaly identification algorithm to corrosion rate data obtained from on-site inspections of oil and gas pipelines may fail to accurately identify anomalies, leading to errors in the pipeline corrosion analysis results based on the pipeline corrosion rate data.
[0006] In view of the above, this application is hereby submitted. Summary of the Invention
[0007] The problem with existing technologies is that the detected corrosion data inside pipelines varies in terms of sample size, data dimensionality, and normality. These differences lead to different applicable anomaly identification algorithms, and it is impossible to clearly identify which internal corrosion data are normal. This invention provides an optimized method for identifying anomaly data of corrosion rate inside natural gas pipelines. It can select the algorithm with the best applicability for identifying anomaly data of corrosion rate inside pipelines. The optimized algorithm can identify anomalies in the massive corrosion rate data continuously generated by the subsequent gathering and transmission pipeline system, ensuring the authenticity of the internal corrosion rate data and laying the foundation for the reliability of corrosion analysis results.
[0008] This invention is achieved through the following technical solution:
[0009] A preferred method for identifying abnormal corrosion rate data in natural gas pipelines includes the following steps:
[0010] S1, collect commonly used anomaly data identification algorithms and obtain more than 5 mainstream anomaly data identification algorithms;
[0011] S2, Collect internal corrosion rate data of pipelines in the same system and identify outliers in the internal corrosion data;
[0012] S3. Determine the sample size, data dimension, and normality of the corrosion rate data in the pipeline, and use the sample size, data dimension, and normality as indicators for algorithm selection.
[0013] S4. The weights W1 for sample size, W2 for data dimension, and W3 for normality are determined using the analytic hierarchy process.
[0014] S5 divides the three indicators of sample size, data dimension and normality into multiple applicability levels. Based on the sample size, data dimension and whether the corrosion rate data in the pipeline meets the normal distribution, quantitative values are assigned to each level.
[0015] S6: Based on the applicability level of each indicator in S1, obtain the corresponding indicator's level score, multiply the weight of each indicator by the corresponding level score, and then add them together to obtain the score of the single algorithm.
[0016] S7. Repeat step S6 to obtain the scores of all algorithms, and select the top-ranked algorithms according to the ratio of scores.
[0017] S8 applies the selected algorithms to the corrosion rate data in the pipeline for anomaly identification, and selects the optimal algorithm based on the anomaly identification rate of the various algorithms.
[0018] Furthermore, in S1, commonly used anomaly detection algorithms include Laida criterion, box plot, absolute median difference, Grubbs criterion, DBSCAN clustering, K-means clustering, local outlier factor, isolated forest, and KNN anomaly detection. The anomaly detection algorithm includes the algorithm's calculation steps, adaptability, and anomaly detection criteria.
[0019] Furthermore, S2 also includes data preprocessing, specifically supplementing or removing invalid data in the dataset, organizing the data by pipeline, and integrating it to obtain a pipeline corrosion rate database. The database includes pipeline name, number of normal data, number of abnormal data, and specific data points.
[0020] Furthermore, in S3, a data volume ≤ 1000 is considered a small sample data volume, and a data volume > 1000 is considered a large sample data volume.
[0021] Furthermore, in S3, the data significance P>0.05 or |kurtosis|≤2|skewness|≤2 indicates that the data follows a normal distribution.
[0022] Furthermore, in S4, the importance of the three indicators—sample size, data dimensionality, and normality—is first assessed, and the assessment criteria are shown in the table below:
[0023]
[0024] Furthermore, in S4, the judgment matrix between the indicators is then determined, as shown in the judgment matrix table below:
[0025] index Sample size Data Dimensions Normality Sample size 1 1 / α 1 / β Data Dimensions α 1 1 / γ Normality β γ 1
[0026] α, β, and γ are quantitative values representing the degree of importance among the indicators;
[0027] Finally, the weights of each indicator are determined using the analytic hierarchy process (AHP). The specific algorithm is as follows:
[0028] (1) The weight values of each evaluation index can be calculated based on the constructed judgment matrix, and the weights of each index can be solved using the square root method:
[0029] The first step is to calculate the nth root of the quantitative score of each indicator in the judgment matrix:
[0030]
[0031] In the formula a ij —This indicates that the comparison result of the i-th indicator with respect to the j-th indicator is in the judgment matrix;
[0032] The second step is to M i Normalization yields the weight values W1, W2, and W3 for each evaluation index, which are then combined to obtain the eigenvector W of the judgment matrix.
[0033]
[0034] The third step is to calculate the largest eigenvalue of the judgment matrix:
[0035]
[0036] (2) To determine whether the matrix is reliable, a consistency check is required. This check is determined by the largest eigenvalue of the matrix and the number of evaluation indicators. The consistency indicator is defined as follows:
[0037]
[0038] In the formula λ max — Determine the largest eigenvalue of the matrix; n — Determine the order of the matrix. When n < 3, the matrix consistency test result is good. When n is large, it is necessary to check C. I Make corrections:
[0039]
[0040] In the formula R I —Correction factor, values are shown in the table below:
[0041] Matrix order 1 2 3 4 5 6 7 8 9 10 <![CDATA[R I ]]> 0 0 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49
[0042] Generally, when C R If the value is less than 0.1, it means that the established judgment matrix meets the one-time test requirement, and the weight values W1, W2, and W3 of each evaluation index are finally obtained.
[0043] Furthermore, in S5, the three indicators of sample size, data dimension, and normality are divided into five applicability levels: best (level score 1), good (level score 0.8), average (level score 0.5), poor (level score 0.2), and very poor (level score 0).
[0044] Furthermore, in S8, the three selected algorithms are applied to the corrosion rate data in the pipeline for anomaly identification. The anomaly identification rates of the three algorithms for all corrosion rate data in the pipeline are analyzed and compared. The anomaly identification rate is the ratio of the number of anomaly data identified by the algorithm to the number of anomaly data in the database. The optimal algorithm with the anomaly identification rate closest to or equal to 1 is selected.
[0045] Compared with the prior art, the present invention has the following advantages and beneficial effects:
[0046] The present invention provides an optimal method for identifying anomaly data of corrosion rate in natural gas pipelines. This method can select the algorithm with the best applicability for identifying anomaly data of corrosion rate in pipelines. The optimal algorithm can identify anomalies in the massive amount of corrosion rate data continuously generated by the subsequent gathering and transmission pipeline system, ensuring the authenticity of the internal corrosion rate data and laying the foundation for the reliability of corrosion analysis results. Attached Figure Description
[0047] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0048] Figure 1 A flowchart illustrating a preferred method for implementing the abnormal corrosion data algorithm provided in this embodiment of the invention;
[0049] Figure 2 A flowchart for the normality test of corrosion rate data in pipelines provided in an embodiment of the present invention. Detailed Implementation
[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments. The illustrative embodiments and descriptions of this invention are only used to explain this invention and are not intended to limit this invention.
[0051] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention.
[0052] Throughout this specification, references to "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "an embodiment," "an example," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described herein, as well as the features of those different embodiments or examples.
[0053] The terminology used throughout this specification consists of those general terms currently widely used in the art in consideration of the functionality of this disclosure; however, these terms may vary depending on the intent, precedent, or new technology of a person skilled in the art. Furthermore, specific terms may be chosen by the applicant, and in such cases, their detailed meanings will be described in the detailed description of this disclosure. Therefore, the terminology used in this specification should not be construed as simple names, but rather based on the meaning of the terms and the overall description of this disclosure.
[0054] This specification uses flowcharts or text to illustrate the operational steps performed according to the embodiments of this application. It should be understood that the operational steps in the embodiments of this application are not necessarily performed precisely in the order described. Instead, as needed, various steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more operations can be removed from these processes.
[0055] In the description of this invention, the terms "center," "length," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "inner," "outer," "circumferential," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this specification and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this specification.
[0056] Example 1
[0057] The present invention provides a preferred method for identifying abnormal corrosion rate data in natural gas pipelines, the process of which is as follows: Figure 1 As shown, the specific steps are as follows:
[0058] (1) Research commonly used anomaly identification algorithms: obtain more than 5 mainstream anomaly identification algorithms, including the calculation steps, applicable situations, anomaly identification standards, etc. These algorithms can be found in journal articles, dissertations and related websites;
[0059] (2) Collect corrosion rate data in pipelines of the same system: Collect corrosion rate data and outliers of some corrosion data in pipelines. Relevant data can be obtained from pipeline detection, corrosion probe, etc. Finally, the data are sorted by pipeline.
[0060] (3) Preprocessing the dataset by supplementing or removing invalid data;
[0061] (4) Establish an algorithm selection index system: Determine the data dimensions of the corrosion rate data in the pipeline through the data processing in step (2); based on... Figure 2 The method shown determines the sample size of the data, defining a data volume of 1000 or less as a small sample size, and vice versa; according to Figure 2 The method shown determines the normality of the data; sample size, data dimensionality, and normality are used as three indicators for algorithm selection.
[0062] (5) Determine the weights of the indicators: The importance of the three indicators is judged as shown in Table 1, and the judgment matrix of the three indicators is shown in Table 2. The weights (W1, W2, W3) of the three indicators of sample size, data dimension and normality are finally determined by the analytic hierarchy process.
[0063] Table 1. Scale of the Judgment Matrix
[0064] The importance of indicator i compared to indicator j Quantized value Indicator i is as important as indicator j 1 Indicator i is slightly more important than indicator j. 3 Indicator i is stronger and more important than indicator j 5 Indicator i is significantly more important than indicator j. 7 Indicator i is extremely more important than indicator j 9 The median of the adjacent judgment between indicator i and indicator j 2、4、6、8
[0065] Table 2 Judgment Matrix
[0066]
[0067] (6) Scoring of indicators: The three indicators of sample size, data dimension and normality are divided into five levels. Based on the sample size, data dimension and whether the corrosion rate data in the pipeline meets the normal distribution, a quantitative value is assigned to each level. The algorithm is most suitable for the data type indicator and is scored as 1 point, good applicability as 0.8 points, average as 0.5 points, poor as 0.2 points, and very poor as 0 points. This is used to quantitatively judge the applicability of the algorithm to the three indicators of sample size, data dimension and normality.
[0068] (7) Formulate a scoring table as shown in Table 3, multiply the weight of each indicator by the corresponding grade score to obtain the individual indicator score (X1, X2, X3), and finally obtain the final score (X) of the individual algorithm, as shown below;
[0069] X = X1 + X2 + X3 (1)
[0070] Table 3 Scoring Table
[0071]
[0072] (8) Complete the comparison of abnormal data identification algorithms: Based on the scoring table, obtain the final score of all algorithms, and select the three algorithms with the highest scores according to the score ratio.
[0073] (9) Establish a pipeline corrosion rate database. Through the collection and preprocessing of corrosion data, the final integrated database includes the pipeline name, the number of normal data (a1, ...), and the number of abnormal data (a...). 11 The relevant information of the data points (x1, x2, x3, ...) and specific data points (x1, x2, x3, ...) are shown in Table 4, which serve as the evaluation information for the selection of anomaly data identification algorithms.
[0074] Table 4. Database Form for Pipeline Corrosion Rate
[0075] pipeline Number of normal internal corrosion rate data Number of abnormal data Data points Pipe A <![CDATA[a1]]> <![CDATA[a 11 ]]> <![CDATA[x1、x2、x3、……]]> …… …… …… ……
[0076] (10) Evaluation of abnormal data identification results: Based on the three algorithms selected by comparison, they were applied to the corrosion rate data in the pipeline to identify abnormal data; an abnormal data identification rate index was established as the evaluation standard, that is, the ratio of the number of abnormal data identified by the algorithm to the number of abnormal data in the database. The closer the ratio is to 1, the better the abnormal data identification rate of the algorithm, indicating that the algorithm is more applicable to the pipeline system; the abnormal data identification rates (a#%, b#%, c#%, ...) of the three algorithms for all corrosion rate data in the pipeline were analyzed and compared, as shown in Table 5. The algorithm with the abnormal data identification rate closest to 1 is the optimal algorithm.
[0077] Table 5 Anomaly detection results of different algorithms
[0078]
[0079] Example 2
[0080] Based on actual pipeline internal corrosion rate data, the preferred method for identifying abnormal internal corrosion rate data in natural gas pipelines, provided in this embodiment of the invention, is as follows:
[0081] (1) The survey found nine anomaly detection algorithms, including Laida criterion, box plot, absolute median difference, Grubbs criterion, DBSCAN clustering, K-means clustering, local outlier factor, isolated forest and KNN anomaly detection, as well as their algorithm calculation steps, adaptability, anomaly identification criteria, etc.
[0082] (2) Complete the collection of corrosion rate data and some corrosion anomaly data in the pipeline;
[0083] (3) Supplement and remove invalid data in the corrosion data;
[0084] (4) The corrosion rate data in the pipeline was determined to be small sample, single (low) dimension, and non-normally distributed;
[0085] (5) Determine the judgment matrix between the indicators, as shown in Table 6 below:
[0086] Table 6 Judgment Matrix
[0087] index Sample size Data Dimensions Normality Sample size 1 2 3 Data Dimensions 0.5 1 2 Normality 0.3333 0.5 1
[0088] The cube root of the product of the quantified values of each index in the judgment matrix is calculated:
[0089] By normalization, the weights of the indicators for sample size, data dimension, and normality are obtained: Sample size weight = 1.8171 / (1.8171+1+0.5501) = 0.5396, Data dimension weight = 1 / (1.8171+1+0.5501) = 0.2970, Normality weight = 0.5501 / (1.8171+1+0.5501) = 0.1634.
[0090] The largest eigenvalue λ of the judgment matrix is calculated. max The value is 3.0088. Since the order of the judgment matrix is 3, the consistency index C is... I = (3.0088-3) / (3-1) = 0.0044, through R I The correction factor is adjusted to obtain the final consistency index C. R =0.0076, which is less than 0.1, indicating that the established judgment matrix meets the one-time test requirement. The final weights of the three indicators, sample size, data dimension, and normality, are 0.5396, 0.297, and 0.1634, respectively.
[0091] (6) For the nine algorithms, the algorithm that is most suitable for small sample size, low-dimensional data or does not require normal distribution of data is given a score of 1 for the corresponding index level, 0.8 for good applicability, 0.5 for average, 0.2 for poor, and 0 for very poor.
[0092] (7) The individual index scores of the nine algorithms were obtained, and the final scores of the nine algorithms were obtained: Box plot got 0.8594 points, K-means clustering got 0.8 points, DBSCAN clustering got 0.8 points, Grubbs criterion got 0.7287 points, local outlier factor got 0.4005 points, Laida criterion and absolute median difference both got 0.4049 points, KNN anomaly detection and isolated forest both got 0.2302 points.
[0093] (8) The three algorithms with the highest scores were selected according to the ratio of high to low scores: box plot, K-means clustering and DBSCAN clustering.
[0094] (9) Retrieve relevant information about pipeline Z from the pipeline corrosion database. There are four outliers in the corrosion rate data of this pipeline, namely data points 0.0239, 0.0502, 0.0331, and 0.2585, as shown in Table 7:
[0095] Table 7. Relevant data for pipeline Z
[0096]
[0097] (10) Three algorithms, namely box plot, K-means clustering and DBSCAN clustering, were applied to the internal corrosion rate data of pipeline Z for anomaly identification. As shown in Table 8, based on the anomaly identification rate, box plot was selected as the optimal algorithm. The optimal algorithm for identifying anomaly internal corrosion rate data of pipeline was completed. The optimal algorithm can be used to identify anomalies in the massive corrosion rate data continuously generated by the subsequent gathering and transportation pipeline system.
[0098] Table 8 Anomaly detection results of different algorithms
[0099]
[0100] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines, characterized in that, Includes the following steps: S1, collect commonly used anomaly detection algorithms and obtain more than 5 anomaly detection algorithms; the commonly used anomaly detection algorithms include Laida criterion, box plot, absolute median difference, Grubbs criterion, DBSCAN clustering, K-means clustering, local outlier factor, isolated forest and KNN anomaly detection, and the anomaly detection algorithm includes the calculation steps, adaptability and anomaly detection criteria of the algorithm; S2, Collect internal corrosion rate data of pipelines in the same system and identify outliers in the internal corrosion data; S3. Determine the sample size, data dimension, and normality of the corrosion rate data in the pipeline, and use the sample size, data dimension, and normality as indicators for algorithm selection. S4. The weights for the sample size are determined using the analytic hierarchy process (AHP). W 1. Weights of data dimensions W 2 and the weight of normality W 3; S5 divides the three indicators of sample size, data dimension and normality into multiple applicability levels. Based on the sample size, data dimension and whether the corrosion rate data in the pipeline meets the normal distribution, quantitative values are assigned to each level. S6: Based on the applicability level of each indicator in S1, obtain the corresponding indicator's level score, multiply the weight of each indicator by the corresponding level score, and then add them together to obtain the score of the single algorithm. S7. Repeat step S6 to obtain the scores of all algorithms, and select the top-ranked algorithms according to the ratio of scores. S8 applies the selected algorithms to the corrosion rate data in the pipeline for anomaly identification, and selects the optimal algorithm based on the anomaly identification rate of the various algorithms.
2. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 1, characterized in that, S2 also includes data preprocessing, specifically supplementing or removing invalid data in the dataset, organizing the data by pipeline, and integrating it to obtain a pipeline corrosion rate database. The database includes pipeline name, number of normal data, number of abnormal data, and specific data points.
3. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 1, characterized in that, In S3, data volume ≤ 1000 is considered small sample data volume, and data volume > 1000 is considered large sample data volume.
4. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 1, characterized in that, In S3, data with significance P>0.05 or |kurtosis|≤2 |skewness|≤2 indicate that the data follows a normal distribution.
5. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 1, characterized in that, In S4, the importance of the three indicators—sample size, data dimensionality, and normality—is first assessed, and the assessment criteria are shown in the table below: 。 6. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 5, characterized in that, In S4, the judgment matrix between the indicators is then determined, as shown in the table below: ; α, β, γ This is a quantitative value representing the degree of importance between indicators; Finally, the weights of each indicator were determined using the analytic hierarchy process (AHP).
7. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 6, characterized in that, The specific algorithms for calculating the weights of each indicator are as follows: (1) The weight values of each evaluation index can be calculated based on the constructed judgment matrix, and the weights of each index can be solved using the square root method: The first step is to calculate the quantified score of each indicator in the judgment matrix. n The right root: (1) In the formula — This indicates that the judgment matrix is the first one. i The first indicator is relative to the first j Comparison results of the indicators; The second step is to Normalization yields the weight values for each evaluation indicator. W 1. W 2. W 3. Combine to obtain the eigenvectors of the judgment matrix. W : (2); The third step is to calculate the largest eigenvalue of the judgment matrix: (3); (2) To determine whether the matrix is reliable, a consistency check is required. This check is determined by the largest eigenvalue of the matrix and the number of evaluation indicators. The consistency indicator is defined as: (4) In the formula λ max —Determine the largest eigenvalue of the matrix; n — Determine the order of a matrix. n When <3, the matrix consistency test results are better. n When ≥3, it is necessary to... C I Make corrections: (5) In the formula R I —Correction factor, values are shown in the table below: ; when C R If the value is less than 0.1, it indicates that the established judgment matrix meets the one-time test requirement, and the final weight values of each evaluation index are obtained. W 1. W 2. W 3.
8. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 1, characterized in that, In S5, the three indicators of sample size, data dimensionality, and normality are divided into five applicability levels: best (level score 1), good (level score 0.8), average (level score 0.5), and poor (level score 0.2). Poor: The grade score is 0.
9. A preferred method for identifying abnormal corrosion rate data in natural gas pipelines according to claim 1, characterized in that, In S8, the three selected algorithms are applied to the corrosion rate data in the pipeline for anomaly identification. The anomaly identification rates of the three algorithms for all corrosion rate data in the pipeline are analyzed and compared. The optimal algorithm with the anomaly identification rate closest to or equal to 1 is selected. The anomaly identification rate is the ratio of the number of anomaly data identified by the algorithm to the number of anomaly data in the database.
Citation Information
Patent Citations
Distinguished access point optimization algorithm in wireless local area network
CN104955058A
Natural gas pipeline corrosion defect typical characteristic extraction method
CN110486630A
