Gas pipe network line patrol district division method and device and storage medium

Through fuzzy comprehensive evaluation and cluster analysis methods, the problem of not taking into account multiple factors in the existing technology of line patrol area division is solved, and a more scientific and efficient line patrol area division and inspection plan generation is achieved, and the gas pipe is improved The scientific nature of online patrol management and the ability to quickly adjust.

CN118965049APending Publication Date: 2024-11-15GUANGZHOU GAS GROUP CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411202301.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing method of gas pipeline patrol area division depends on human experience and cannot comprehensively consider a variety of factors, such as the length of gas pipeline network in the area, key inspection areas, accident risk level and old pipeline ratio, resulting in poor workload of inspection personnel. Balanced and it is difficult to quickly adjust the plan when personnel or area changes.

Method used

The fuzzy comprehensive evaluation and cluster analysis method is adopted, and the evaluation value of each inspection point is calculated by setting up the latitude and longitude coordinates, attribute preprocessing and fuzzy comprehensive evaluation model of the inspection point, and the clustering model is established to divide the inspection line area, considering multiple and automatically generate inspection plans.

Benefits of technology

It has achieved a more scientific division of line patrol areas, and can quickly adjust the inspection areas and plans, improve the scientificity and efficiency of line patrol management, and ensure the balance and adaptability of inspection tasks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118965049A_ABST
    Figure CN118965049A_ABST
Patent Text Reader

Abstract

The invention discloses a gas pipe network line patrol district division method and device and a storage medium, and the method comprises the steps: building an evaluation model # imgabs0 # during the district division, considering a plurality of factors affecting the gas pipe network line patrol, and not carrying out the clustering analysis only according to the distance of each patrol point at the geographic position. The evaluation value # imgabs1 # of each inspection point is calculated by substituting data such as accident risk level, gas pipeline age limit, corrosion leakage condition and key area distance, so that multi-dimensional quantitative analysis and evaluation are realized on data characteristics contained in each inspection point. Then, the positions and the number of divided areas are designated by a user, a distance measurement operator # imgabs2 #, a cost function # imgabs3 # and a contour coefficient # imgabs4 # are set, a # imgabs5 # clustering model is constructed and substituted into an evaluation value # imgabs6 # and latitude and longitude coordinates of each inspection point for solving, and finally, an optimal partition and inspection scheme is provided for the user. While the scientific management level of the pipe network line patrol district is improved, the effects of quickly dividing the inspection area and automatically generating the patrol scheme are realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of gas pipeline network laying, and in particular to a gas pipeline network patrol area division method based on fuzzy evaluation and cluster analysis. Background Art

[0002] Regularly conducting gas pipeline inspections is one of the most important safety assurance tasks for gas companies. Staff members complete the gas pipeline inspections in the area they are responsible for within the specified time by walking, riding bicycles, and driving inspection vehicles. A large amount of manpower and material resources are required to complete the full coverage inspection of the pipelines in the jurisdiction. At present, most gas companies divide the patrol areas on a large scale according to administrative regions, and then divide them into small areas based on the number of personnel that can be deployed and the mode of travel, and according to long-term management experience and the familiarity of the patrol staff with the geographical location. The overall model can be summarized as follows: first, the patrol area is divided according to experience and is responsible for by specific personnel, and then patrol points are set up according to the inspection standards in the designated patrol area where the gas pipeline is buried, so as to determine the work area and task volume of the patrol personnel.

[0003] By using appropriate algorithms, various factors that affect the setting of patrol areas can be taken into account, and the area division and patrol setting scheme can be quickly calculated by computer. The area division problem can be analyzed and modeled by target optimization, so that a planning algorithm or heuristic algorithm can be used to quickly solve the feasible domain with high enough accuracy. Machine learning or mathematical statistics methods can also be used to model and solve, such as clustering, greedy and regression methods. By using existing data for training and summarizing the experience of reasonable division, the model can learn how to comprehensively consider the various factors involved in the area division problem and realize the rapid division of gas inspection areas.

[0004] However, since the existing patrol work mode divides the area according to human experience, and adopts the order of dividing the area first and then setting the inspection points, there will be the following problems: 1. When dividing the inspection area, the area can only be roughly judged, and multiple factors cannot be comprehensively considered. Such as the length of the gas pipeline network in the area, whether there are key inspection areas such as schools and hospitals in the area, the risk level of gas pipeline accidents, and the proportion of old gas pipelines. 2. Since the inspection points are not set first and then the areas are divided, it is impossible to know the distance of pipeline inspection in each area, resulting in different workloads for patrol personnel. In addition, there are large differences in road conditions in each area, and there may be a phenomenon that the area is small but the inspection task is large. 3. Once there is an increase or decrease in personnel or a temporary change in the inspection area, it will cause the management personnel to spend a lot of time and energy to readjust and formulate the location of the inspection area and the inspection point, and it is impossible to quickly generate a gas pipeline network patrol plan based on the new situation. Summary of the invention

[0005] A gas network patrol area division method proposed in the present invention can at least solve one of the technical problems in the background technology.

[0006] To achieve the above object, the present invention adopts the following technical solutions: A gas pipeline network patrol area division method uses the results of fuzzy comprehensive evaluation as the basis for clustering to generate a regional division plan. The specific steps are as follows: S1. Set up key gas inspection areas, general inspection point locations, and supplementary inspection point locations; S2. Obtain the latitude and longitude coordinates of the key gas inspection area , the latitude and longitude coordinates of the locations of ordinary inspection points and supplementary inspection points ; S3, supplement various attributes of inspection points, and conduct quantitative analysis, dimension conversion and data preprocessing on attribute values; S4. Build a fuzzy comprehensive evaluation model ; Combine the various attributes of the inspection points to obtain the evaluation value of each inspection point i ; S5. Fuzzy comprehensive evaluation model Output inspection review value i As the basis for distance calculation, establish The clustering model is used to calculate the area division results.

[0007] Furthermore, the method for obtaining the longitude and latitude coordinates of the key inspection area in S2 of the present invention is as follows: Convert the key inspection area into one or more longitude and latitude coordinates. If the area is larger than a 300*300m square area, divide the key inspection area into multiple square areas and obtain the longitude and latitude coordinates of the center of each square. If the area of ​​the key inspection area is less than or equal to a 300*300m square area, the longitude and latitude coordinates of the geometric center of the area are directly taken.

[0008] Furthermore, the method for obtaining the latitude and longitude coordinates of the inspection points in S2 of the present invention is as follows: S21. Set the minimum distance between inspection points , set up common inspection points according to the minimum interval distance, and obtain the latitude and longitude coordinates of the inspection points ; S22. Take the gas pipeline valves, tees, pressure regulating equipment and damaged points as supplementary inspection points to obtain the longitude and latitude coordinates .

[0009] Furthermore, the method for preprocessing attribute value data in S3 of the present invention is specifically as follows: S31. Risk level for non-numerical data accidents i, The values ​​are low risk, medium risk and high risk, and the categorical variables are converted into numerical data using One-Hot Encoding or Label Encoding. S32. For data with missing data, if the missing value is less than 30%, the mean value is used to fill the missing value. If the missing value is greater than or equal to 30%, the K-nearest neighbor algorithm is used to predict the missing value with the existing data. S33. For data with different dimensions, the accident risk level risk i , Pipeline age at inspection point i , the degree of corrosion and leakage of pipelines at inspection points i The time required to travel with the minimum walking distance point_walk , use the Min-Max normalization algorithm to scale the data to the range of [0,1].

[0010] Furthermore, the fuzzy comprehensive evaluation model in S4 of the present invention The construction method is as follows: S41. Set factor sets for inspection points based on existing data , Comments , the specific format is as follows: ;

[0011] in: = Time to pass through inspection points and key areas , = Inspection point attributes , = Inspection point attributes , = Inspection point attributes ; In Including walking, cycling and driving time , and ; =Evaluated as excellent, =Evaluated as good, =Evaluated as average, =Evaluation is poor; S42. Construct an evaluation matrix for each inspection point based on the comment set V , the matrix format is as follows: ,

[0012]

[0013] in It is the evaluation value of a single factor. A factor can only have one evaluation; S43. Combined with the setting method of evaluation matrix R, construct a fuzzy comprehensive evaluation model :

[0014] Among them: W is the weight matrix, R is the evaluation matrix, Evaluate is the fuzzy decision vector, n is the number of evaluation indicators, which is consistent with the number in the comment set V; m is the number of evaluation objects, which is consistent with the number of inspection points; S44, The settings of each weight in the weight matrix need to be checked for consistency. The checking method is as follows:

[0015] Among them: RI is the average random consistency index, which depends on the matrix order; if CR<0.1, the weight setting is reasonable; S45. Constructing fuzzy synthesis operator , and is constructed as follows: .

[0016] Furthermore, the evaluation value value in S4 of the present invention i The acquisition method is as follows: Step 1: Substitute the pre-processed data obtained in S3 to calculate the fuzzy evaluation vector of each inspection point (including ordinary inspection points and supplementary inspection points) ; Step 2: In order to comprehensively consider the impact of various attributes on the division area when calculating the distance between each inspection point, a weighted average operator is constructed to calculate the value i ; The formula of the weighted average operator is expressed as follows:

[0017] in: for The element value of is the corresponding score of each evaluation index, the value of n and the number of evaluation indicators, and The number of elements is consistent; Step 3: The fuzzy evaluation vector of each inspection point Convert to evaluation value i .

[0018] Furthermore, the method for building the K-Means-Evaluate clustering model in S5 of the present invention is as follows: S51. Setting data sample set Distance metric operator ; The calculation is as follows:

[0019]

[0020] S52, judging whether the main parameters have reached the optimal value by minimizing a cost function; Setting the cost function , the formula is as follows:

[0021] Where K is the number of cluster centers, is the set of inspection points classified into the same category as the cluster center k, is the location representation of cluster center k; S53, for accurate assessment The clustering effect of the model, in addition to using the cost function In addition to representing the dispersion of clusters, set the silhouette coefficient Measure the density and sparsity of clustering; The specific formula is as follows:

[0022] Where: N is the number of inspection points in the current area, M is the number of inspection points in the nearest adjacent area, inside and outside both represent average distances; S54, Establishment Clustering model solving method.

[0023] Furthermore, in S54 of the present invention, The clustering model solution method is as follows: S541. Set the expected number of cluster centers, i.e. the number of partition areas. ; S542, Random Selection inspection points and use them as the proposed cluster centers to calculate their cost functions ,choose The solutions with the smallest values ​​are selected as the determined cluster centers; S543, calculate the distance between all inspection points and cluster centers under several schemes in S542 For each inspection point, divide it into the nearest cluster center to form small areas and calculate its silhouette coefficient ; S544. Comparison of silhouette coefficients under different partitioning schemes The largest value is selected as the final partitioning scheme; if the partitioning effect is not good, reselect value, and in Repeat the above steps within the range of , so as to obtain the optimal partition of the inspection plan.

[0024] On the other hand, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0025] On the other hand, the present invention further discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0026] The main function of the present invention is to combine the fuzzy comprehensive evaluation algorithm with the clustering algorithm, and divide the gas pipeline network inspection area through quantitative analysis, data processing, model construction and iterative calculation, thereby replacing the traditional practice of direct zoning based on human experience, and being able to specify the number and location of the partitions according to management experience. While improving the scientific management level of the pipeline network patrol area, it can achieve the effect of quickly dividing the inspection area and automatically generating the inspection plan.

[0027] In order to comprehensively consider multiple influencing factors when dividing the area, rather than just based on the geographical distance of each inspection point, a fuzzy comprehensive evaluation model is constructed. By substituting data such as accident risk level, pipeline age, corrosion leakage, and distance to key areas (the distance between the inspection point and the adjacent key area can be measured by the travel time of three different modes of travel), the evaluation value of each inspection point is calculated. , which is used to quantitatively analyze and evaluate the data characteristics contained in each inspection point from multiple dimensions.

[0028] Then the user specifies the location and number of the divided areas, and sets the distance metric operator , cost function and silhouette coefficient , build Clustering model and substitute the evaluation value of each inspection point Solve with longitude and latitude coordinates to get the optimal zoning and inspection plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1It is a flowchart of the present invention. DETAILED DESCRIPTION

[0030] To make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them.

[0031] like Figure 1 As shown in the figure, a gas pipeline patrol area division method based on fuzzy evaluation and cluster analysis is provided. This example takes the urban gas pipeline patrol area division as an example, and specifically includes the following steps: Step 1: Set the key gas inspection areas, general inspection point locations, and supplementary inspection point locations; The key gas inspection areas include the geographical locations of key areas such as hospitals, schools, military regions, government units and residential areas. These areas have high inspection requirements and great impact from gas accidents, which have an impact on the evaluation of inspection points (the closer the inspection point is to the key area, the greater the impact); ordinary inspection points are set above the roads where gas pipelines are buried, and additional inspection points are set according to the pipeline structure.

[0032] S2. Obtain the latitude and longitude coordinates of the key gas inspection area , the latitude and longitude coordinates of the locations of ordinary inspection points and supplementary inspection points ; The method for obtaining the latitude and longitude coordinates of the key inspection area is as follows: Convert the key inspection area into one or more longitude and latitude coordinates. If the area is larger than a 300*300m square area, divide the key inspection area into multiple square areas and obtain the longitude and latitude coordinates of the center of each square. If the area of ​​the key inspection area is less than or equal to a 300*300m square area, the longitude and latitude coordinates of the geometric center of the area are directly taken.

[0033] The method for obtaining the latitude and longitude coordinates of the inspection point includes: S21. Set the minimum distance between inspection points , set up common inspection points according to the minimum interval distance, and obtain the latitude and longitude coordinates of the inspection points ; S22. Take the gas pipeline valve, tee and damaged point as supplementary inspection points to obtain the longitude and latitude coordinates .

[0034] S3, supplement various attributes of inspection points, and conduct quantitative analysis, dimension conversion and data preprocessing on attribute values; The properties of the inspection points are as follows: Call the path planning algorithm interface, substitute the latitude and longitude coordinates of the inspection points obtained in S2, and calculate the travel time between each inspection point and the key area according to the three modes of walking, cycling and driving, which are recorded as , and ; Calculate the minimum distance between walking, cycling and driving The travel time is recorded as , and , it can avoid repeatedly calculating the travel time between inspection points; The travel time between supplementary inspection points such as valves, tees and damaged points and a common inspection point (usually the common inspection points in the west and south directions) is calculated according to the three modes of walking, cycling and driving, and recorded as , and ; The path planning algorithm interface refers to existing products sold on the market, such as Baidu Map Open Platform API and Amap API, or open source products.

[0035] Since the supplementary inspection point is located between two ordinary inspection points, the travel time between the supplementary inspection point and another ordinary inspection point (the ordinary inspection point in the east and north directions) can be recorded as , and ; According to the accident risk level of the pipeline (the higher the risk, the more likely the pipeline is to have an accident), add the attributes of the general inspection points and the supplementary inspection points risk i , according to the age of the pipeline, add the attributes of the general inspection point and the supplementary inspection point year i , according to the corrosion and leakage of the pipeline, add the properties of the general inspection points and the supplementary inspection points i .

[0036] The method for preprocessing the attribute value data is specifically as follows: S31. For non-numerical data, such as accident risk level risk i (values ​​are low risk, medium risk, and high risk), use One-Hot Encoding or Label Encoding to convert categorical variables into numerical data; S32. For data with missing data, if the missing value is less than 30%, the mean value is used to fill the missing value. If the missing value is more than 30%, the K-nearest neighbor algorithm is used to predict the missing value with the existing data. S33、For data with different dimensions, such as riski 、year i 、corrode i and time point_walk etc., using the Min-Max normalization algorithm to scale the data to the range of [0,1].

[0037] S4. Build a fuzzy comprehensive evaluation model ; Combine the various attribute values ​​of the inspection points to obtain the evaluation value of each inspection point i ; Since each inspection point contains more than one attribute, in order to calculate the distance between different inspection points when using the clustering algorithm (the inspection points with similar attributes are closer), a fuzzy comprehensive evaluation model is established based on the data structure obtained in S3. , use the evaluation value i The difference between the sample and the centroid is calculated; The fuzzy comprehensive evaluation model The construction method is as follows: S41. Set factor sets for inspection points based on existing data , Comments , the specific format is as follows: ;

[0038] in: = Time to pass through inspection points and key areas , = Inspection point attributes , = Inspection point attributes , = Inspection point attributes ; In Including walking, cycling and driving time , and ; =Evaluated as excellent, =Evaluated as good, =Evaluated as average, =Evaluation is poor; S42. Construct an evaluation matrix for each inspection point based on the comment set V , the matrix format is as follows: ,

[0039]

[0040] in is the evaluation value of each factor. A factor can only have one evaluation; S43. Construct a fuzzy comprehensive evaluation model based on the evaluation matrix ;

[0041] Among them: W is the weight matrix, R is the evaluation matrix, Evaluate is the fuzzy decision vector, n is the number of evaluation indicators, which is consistent with the number in the comment set V; m is the number of evaluation objects, which is consistent with the number of inspection points; S44, The settings of each weight in the weight matrix need to be checked for consistency. The checking method is as follows:

[0042] Among them: RI is the average random consistency index, which depends on the matrix order; if CR<0.1, the weight setting is reasonable; S45. Constructing fuzzy synthesis operator , and is constructed as follows:

[0043] So far, the fuzzy comprehensive evaluation model Build completed.

[0044] The method for obtaining the evaluation value valuei is as follows: Step 1: Substitute the pre-processed data obtained in S3 to obtain the fuzzy evaluation vector of each inspection point, including ordinary inspection points and supplementary inspection points. ; Step 2: In order to comprehensively consider the impact of various attributes on the division area when calculating the distance between each inspection point, a weighted average operator is constructed to calculate the value i ; The formula of the weighted average operator is expressed as follows:

[0045] in: for The element value of is the corresponding score of each evaluation index, the value of n and the number of evaluation indicators, and The number of elements is consistent; Step 3: The fuzzy evaluation vector of each inspection point Convert to evaluation value i ; Evaluation value i The meaning of is to construct a variable to represent the multiple attributes of the inspection point. Inspection points with similar attributes are more likely to be divided into the same area.

[0046] S5. Fuzzy comprehensive evaluation model Output inspection review value i As the basis for distance calculation, establish The clustering model is used to calculate the area division results.

[0047] The K-Means-Evaluate clustering model construction method is as follows: S51. Setting data sample set Distance metric operator , The calculation is as follows:

[0048]

[0049] S52, judging whether the main parameters have reached the optimal value by minimizing a cost function; Cost Function The formula is as follows:

[0050] Where K is the number of cluster centers, is the set of inspection points classified into the same category as the cluster center k, is the location representation of cluster center k; Clustering is an unsupervised machine learning algorithm whose main parameters include the location of the cluster center and the inspection points assigned to each cluster.

[0051] S53, for accurate assessment The clustering effect of the model, in addition to using the cost function In addition to representing the dispersion of clusters, set the silhouette coefficient The smaller the difference between inspection points in the same area, the greater the difference between inspection points in different areas, and the greater the silhouette coefficient. The larger the value, the formula is as follows:

[0052] Where: N is the number of inspection points in the current area, M is the number of inspection points in the nearest adjacent area, inside and outside both represent average distances; S54, Establishment Clustering model solving method.

[0053] Said The clustering model solution method is as follows: S541. Set the expected number of cluster centers, i.e. the number of partition areas. ; S542, Random Selection Inspection points are selected as the proposed cluster centers, and the cost function is calculated and selected. ,choose The solutions with the smallest values ​​are selected as the determined cluster centers; Randomly selecting cluster centers may result in Converge to a local optimum rather than approaching the optimum globally. Therefore, the initial setting Small areas evenly distributed in the jurisdiction, try to use several inspection points in the center of the area as cluster centers, and then calculate and select the cost function ,choose The solutions with the smallest values ​​are selected as the real cluster centers. S543, calculate the distance between all inspection points and cluster centers under several schemes in S542 For each inspection point, divide it into the nearest cluster center to form small areas and calculate its silhouette coefficient ; S544. Comparison of silhouette coefficients under different partitioning schemes The largest value is selected as the final partitioning scheme. If the partitioning effect is not good, reselect value, and in Repeat the above steps within the range of , so as to obtain the optimal partition of the inspection plan.

[0054] In summary, the present invention combines fuzzy comprehensive evaluation algorithm and clustering algorithm to divide the gas pipeline network inspection area through quantitative analysis, data processing, model construction and iterative calculation, thereby replacing the traditional practice of direct zoning based on human experience, and can specify the number and location of partitions according to management experience. While improving the scientific management level of pipeline network patrol areas, it can achieve the effect of quickly dividing inspection areas and automatically generating inspection plans.

[0055] In order to comprehensively consider multiple influencing factors when dividing the area, rather than just based on the geographical distance of each inspection point, a fuzzy comprehensive evaluation model is constructed. By substituting data such as accident risk level, pipeline age, corrosion leakage, and distance to key areas (the distance between the inspection point and the adjacent key area can be measured by the travel time of three different modes of travel), the evaluation value of each inspection point is calculated. , which is used to quantitatively analyze and evaluate the data characteristics contained in each inspection point from multiple dimensions.

[0056] Then the user specifies the location and number of the divided areas, and sets the distance metric operator , cost function and silhouette coefficient , build Clustering model and substitute the evaluation value of each inspection point Solve with longitude and latitude coordinates to get the optimal zoning and inspection plan.

[0057] On the other hand, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0058] On the other hand, the present invention further discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0059] In another embodiment provided in the present application, a computer program product comprising instructions is also provided, which, when executed on a computer, enables the computer to execute any of the mobile source emission prediction methods based on time series feature migration in the above-mentioned embodiments.

[0060] It is understandable that the system, device and storage medium provided in the embodiments of the present invention correspond to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts in the above methods.

[0061] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be from a website, a computer, or a server.

Claims

1. A method for dividing gas pipeline network patrol areas, characterized in that: The results of fuzzy comprehensive evaluation are used as the basis for clustering to generate a regional division plan. The specific steps are as follows: S1. Set up key gas inspection areas, general inspection point locations, and supplementary inspection point locations; S2. Obtain the latitude and longitude coordinates of the key gas inspection area , the latitude and longitude coordinates of the locations of ordinary inspection points and supplementary inspection points ; S3, supplement various attributes of inspection points, and conduct quantitative analysis, dimension conversion and data preprocessing on attribute values; S4. Build a fuzzy comprehensive evaluation model ; Combine the various attributes of the inspection points to obtain the evaluation value of each inspection point i ; S5. Fuzzy comprehensive evaluation model Output inspection review value i As the basis for distance calculation, establish The clustering model is used to calculate the area division results.

2. The method for dividing gas pipeline network patrol areas according to claim 1 is characterized in that: The method for obtaining the latitude and longitude coordinates of the key inspection area in S2 is as follows: Convert the key inspection area into one or more longitude and latitude coordinates. If the area is larger than a 300*300m square area, divide the key inspection area into multiple square areas and obtain the longitude and latitude coordinates of the center of each square. If the area of ​​the key inspection area is less than or equal to a 300*300m square area, the longitude and latitude coordinates of the geometric center of the area are directly taken.

3. The method for dividing gas pipeline network patrol areas according to claim 1 is characterized in that: The method for obtaining the latitude and longitude coordinates of the inspection points in S2 is as follows: S21. Set the minimum distance between inspection points , set up common inspection points according to the minimum interval distance, and obtain the latitude and longitude coordinates of the inspection points ; S22. Take the gas pipeline valves, tees, pressure regulating equipment and damaged points as supplementary inspection points to obtain the longitude and latitude coordinates .

4. The method for dividing gas pipeline network patrol areas according to claim 1 is characterized in that: The method for preprocessing attribute value data in S3 is as follows: S31. Risk level for non-numerical data accidents i, The values ​​are low risk, medium risk and high risk, and the categorical variables are converted into numerical data using One-Hot Encoding or Label Encoding. S32. For data with missing data, if the missing value is less than 30%, the mean value is used to fill the missing value. If the missing value is greater than or equal to 30%, the K-nearest neighbor algorithm is used to predict the missing value with the existing data. S33. For data with different dimensions, the accident risk level risk i , Pipeline age at inspection point i , the degree of corrosion and leakage of pipelines at inspection points i The time required to travel with the minimum walking distance point_walk , use the Min-Max normalization algorithm to scale the data to the range of [0,1].

5. The method for dividing gas pipeline network patrol areas according to claim 1 is characterized in that: The fuzzy comprehensive evaluation model in S4 The construction method is as follows: S41. Set factor sets for inspection points based on existing data , Comments , the specific format is as follows: ; in: = Time to pass through inspection points and key areas , = Inspection point attributes , = Inspection point attributes , = Inspection point attributes ; In Including walking, cycling and driving time , and ; =Evaluated as excellent, =Evaluated as good, =Evaluated as average, =Evaluation is poor; S42. Construct an evaluation matrix for each inspection point based on the comment set V , the matrix format is as follows: , in It is the evaluation value of a single factor. A factor can only have one evaluation; S43. Combined with the setting method of evaluation matrix R, construct a fuzzy comprehensive evaluation model : Among them: W is the weight matrix, R is the evaluation matrix, Evaluate is the fuzzy decision vector, n is the number of evaluation indicators, which is consistent with the number in the comment set V; m is the number of evaluation objects, which is consistent with the number of inspection points; S44, The settings of each weight in the weight matrix need to be checked for consistency. The checking method is as follows: Among them: RI is the average random consistency index, which depends on the matrix order; if CR<0.1, the weight setting is reasonable; S45. Constructing fuzzy synthesis operators , and is constructed as follows: 。 6. The method for dividing gas pipeline network patrol areas according to claim 1, characterized in that: The evaluation value in S4 i The acquisition method is as follows: Step 1: Substitute the pre-processed data obtained in S3 to obtain the fuzzy evaluation vector of each inspection point, including ordinary inspection points and supplementary inspection points. ; Step 2: In order to comprehensively consider the impact of various attributes on the division area when calculating the distance between each inspection point, a weighted average operator is constructed to calculate the value i ; The formula of the weighted average operator is expressed as follows: in: for The element value of is the corresponding score of each evaluation index, the value of n and the number of evaluation indicators, and The number of elements is consistent; Step 3: The fuzzy evaluation vector of each inspection point Convert to evaluation value i .

7. The method for dividing gas pipeline network patrol areas according to claim 1, characterized in that: The method for building the K-Means-Evaluate clustering model in S5 is as follows: S51. Setting data sample set Distance metric operator ; The calculation is as follows: S52, judging whether the main parameters have reached the optimal value by minimizing a cost function; Setting the cost function , the formula is as follows: Where K is the number of cluster centers, is the set of inspection points classified into the same category as the cluster center k, is the location representation of cluster center k; S53, for accurate assessment The clustering effect of the model, in addition to using the cost function In addition to representing the dispersion of clusters, set the silhouette coefficient Measure the density and sparsity of clustering; The specific formula is as follows: Where: N is the number of inspection points in the current area, M is the number of inspection points in the nearest adjacent area, inside and outside both represent average distances; S54, Establishment Clustering model solving method.

8. The method for dividing gas pipeline network patrol areas according to claim 7 is characterized in that: The S54 The clustering model solution method is as follows: S541. Set the expected number of cluster centers, i.e. the number of partition areas. ; S542, Random Selection inspection points and use them as the proposed cluster centers to calculate their cost functions ,choose The solutions with the smallest values ​​are selected as the determined cluster centers; S543, calculate the distance between all inspection points and cluster centers under several schemes in S542 For each inspection point, divide it into the nearest cluster center to form small areas and calculate its silhouette coefficient ; S544. Comparison of silhouette coefficients under different partitioning schemes The largest value is selected and used as the final partitioning scheme. If the partitioning effect is not good, reselect value, and in Repeat the above steps within the range of , so as to obtain the optimal partition of the inspection plan.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 8.

10. A computer device comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the computer program is executed by the processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 8.