Power distribution network grid division and whole process management and control method based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based power distribution network grid division method and a full-process management and control method, and the method comprises the steps: obtaining a winding degree index of a cable channel of a first region in a target power distribution network grid division region, determining a target transformer substation, and determining an actual power supply range boundary with the target transformer substation as a center, calculating and screening in the actual power supply range boundary to obtain a plurality of intermediate joint installation positions of the cable in the actual power supply range; determining the evaluation score of each intermediate joint installation position and the distance from each intermediate joint installation position to the actual power supply boundary as a to-be-optimized position; establishing a polar coordinate sampling network according to the to-be-optimized position to determine a final optimized position; and according to the actual power supply range boundary and the final optimized position, dividing the power supply area of the power distribution network to obtain a final result of grid division of the power distribution network. Through the division method, the reliability of the grid division result of the power distribution network is improved, and the operation efficiency of the power distribution network is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network planning, and in particular to a distribution network grid division and full-process control method based on artificial intelligence. Background Art

[0002] In the process of gridding distribution networks in mountainous and hilly areas, the degree of winding of cable trenches has a significant impact on the actual power supply distance. Due to the large terrain undulations, it is difficult to lay cables along a straight path, resulting in the actual power supply distance being much greater than the straight-line distance from the substation to the load center. This difference makes the mapping relationship between the power supply radius within the grid and the substation layout more complicated and difficult to obtain directly through geometric calculations. At the same time, as the weak link in the cable line, the reasonable layout of the cable intermediate joint is the key to ensuring power supply reliability.

[0003] However, in mountainous and hilly areas, the choice of cable intermediate joint layout location is subject to many restrictions due to terrain conditions. How to take into account the electrical performance of the joint location and evaluate the rationality of its layout while meeting the construction conditions is an issue that needs further study. In addition, the above-mentioned evaluation results of the rationality of the cable trench winding degree and the layout location of the cable intermediate joint have important guiding significance for the reasonable delineation of the distribution network grid boundary. However, the delineation of the grid boundary needs to comprehensively consider multiple factors such as the actual power supply distance, cable line direction, and joint layout, in order to achieve the optimization of grid division while ensuring the reliability of power supply and improve the overall operation efficiency of the distribution network. Summary of the invention

[0004] In order to solve the above technical problems, an embodiment of the present invention provides a distribution network grid division and full-process control method based on artificial intelligence to solve the technical problems in the prior art that the distribution network grid division results are of low reliability and reduce the overall operating efficiency of the distribution network.

[0005] A first aspect of an embodiment of the present invention provides a distribution network grid division method based on artificial intelligence, the method comprising: Obtaining terrain characteristic parameters and power load distribution of the first area within the target distribution network grid division area, performing quantitative analysis on the terrain characteristic parameters, obtaining a winding degree index of the first area, obtaining terrain characteristics according to the winding degree index, and obtaining a cable channel path according to the minimum curvature principle and the terrain characteristics; Determine multiple grid nodes according to the distribution of power load, optimize each grid node, obtain multiple substation locations to be laid out, determine the target substation according to the locations of each substation, obtain the optimal path of the cable trench path according to the target substation, correct the optimal path according to the winding degree index, and obtain the actual power supply range boundary of the target substation; Based on the cable load value and the winding index in the actual power supply range boundary, the cable segmentation area is obtained, and the cable segmentation area is calculated using a dynamic programming algorithm to obtain multiple alternative joint positions of the cables. According to the temperature rise curve along the cable, the alternative joint positions of each cable are screened to obtain multiple intermediate joint installation positions of the cables within the actual power supply range; The annual maintenance cost, terrain restriction data and construction cost data of each intermediate joint installation position are calculated by using the hierarchical analysis method to obtain the evaluation score of each intermediate joint installation position, and the distance from each intermediate joint installation position to the actual power supply boundary is calculated to obtain multiple distance values, and the intermediate joint installation position with a distance value less than a preset distance value and an evaluation score less than a preset evaluation value is determined as a position to be optimized; Establish a polar coordinate sampling network according to the position to be optimized, determine the sampling point positions in the polar coordinate sampling network, calculate the power supply distance and voltage drop amplitude of each sampling point position, select the sampling point position whose voltage drop amplitude meets the preset conditions as the optimized position, and obtain the final optimized position according to the load density distribution within the actual power supply range boundary; According to the actual power supply range boundary and the final optimized position, the power supply area of ​​the distribution network is divided to obtain the initial grid division result. The grid boundary is corrected based on the initial grid division result to obtain the final result of the distribution network grid division.

[0006] In a possible implementation manner of the first aspect, a terrain characteristic parameter is quantitatively analyzed to obtain a meandering degree index of a cable channel in the first area, including: According to the terrain characteristic parameters, the height difference ratio between adjacent sampling points in the first area is calculated, and the slope of the sampling point is obtained according to the height difference ratio; Calculate the contour line density distribution in the first area, sample the terrain in the first area at preset intervals based on the contour line density distribution, obtain cross-sectional data, and obtain a winding degree index based on the cross-sectional data and the straight-line distance corresponding to the cross-sectional data.

[0007] In a possible implementation of the first aspect, determining a plurality of grid nodes according to a distribution of power loads, optimizing each grid node, and obtaining a substation location to be laid out includes: A load density index is calculated according to the distribution of power load, and a normalized load density distribution curve is obtained according to the load density index and a load density value greater than a preset density value in the first area; Determine the load center coordinate point according to the normalized load density distribution curve, construct a boundary range according to the load centerline coordinate point and the power point position in the first area, and set grid nodes at a preset interval within the boundary range to obtain a plurality of grid nodes; Each grid node is screened according to a preset power supply radius to obtain a plurality of screened grid nodes, and a simulated annealing algorithm is used to optimize each screened grid node to obtain a plurality of substation locations to be laid out.

[0008] In a possible implementation of the first aspect, according to the target substation, an optimal path of the cable trench path is obtained, and the optimal path is corrected according to a winding degree index to obtain an actual power supply range boundary of the target substation, including: Calculate the impedance value along the grid node to be searched according to the terrain slope data and the reference impedance value of the target substation; A depth-first algorithm is used to traverse the paths of the grid nodes whose impedance values ​​along the path are less than a preset impedance value threshold, and a traversal path set is obtained; The actual length of each path in the traversed path set is calculated with the straight-line distance corresponding to each path to obtain the meandering index of each path, and the coordinate point sequence of the path whose meandering index is less than the preset meandering index is constructed to obtain multiple power supply path combinations; Based on the impedance value and the length of each path, the total impedance value of each power supply path combination is obtained, and the power supply path combination with a total impedance value less than a preset total impedance value threshold is selected as the actual power supply range boundary.

[0009] In a possible implementation of the first aspect, obtaining a cable segment area based on a cable load value and a winding degree index in an actual power supply range boundary includes: The current load value within the boundary of the actual power supply range is obtained, and the voltage drop is obtained according to the current load value and the winding degree index. If the voltage drop is greater than the preset drop threshold, the path within the actual power supply range is segmented to obtain the cable segment area.

[0010] In a possible implementation of the first aspect, a dynamic programming algorithm is used to calculate the cable segment area to obtain multiple candidate joint positions, and the candidate joint positions of each cable are screened according to the temperature rise curve along the cable to obtain multiple intermediate joint installation positions of the cable within the actual power supply range, including: According to the cable segment area and the set search depth threshold, a dynamic programming algorithm is used to determine multiple candidate joint positions within a preset spacing range; Collect conductor temperature data at each candidate joint location; Perform stress analysis on each candidate joint position to obtain the maximum shear stress distribution value of each candidate joint position, determine the stress limit area according to the maximum shear stress distribution value, and select the candidate joint position to be arranged through the stress limit area; If the conductor temperature data corresponding to the alternative joint position to be arranged is less than the preset temperature threshold, and the maximum shear stress distribution value corresponding to the alternative joint position to be arranged is less than the preset material strength, the alternative joint position to be arranged is determined to be the intermediate joint installation position.

[0011] In a possible implementation of the first aspect, the annual maintenance cost, terrain restriction data, and construction cost data of each intermediate joint installation location are calculated using a hierarchy analysis method to obtain an evaluation score for each intermediate joint installation location, including: A judgment matrix for each intermediate joint installation location is established based on the annual maintenance cost value, terrain restriction data, and construction cost data of each intermediate joint installation location; Based on the judgment matrix of each intermediate joint installation position, the scale value is used for calculation to obtain the characteristic vector of each intermediate joint installation position, and the characteristic vector is normalized to obtain the weight vector. If the consistency ratio of the weight vector is less than the preset ratio, the terrain restriction weight value, maintenance cost weight value, and construction cost weight value are extracted from the weight vector. The terrain restriction weight value, maintenance cost weight value and construction cost weight value are weighted together with the corresponding annual maintenance cost value, terrain restriction data and construction cost data to obtain the evaluation score of each intermediate joint installation position.

[0012] In a possible implementation of the first aspect, obtaining a final optimized position according to a load density distribution within a boundary of an actual power supply range includes: According to the load density distribution within the actual power supply range, a location where the load density is greater than a preset load density threshold is selected as a branch connection point; A new position combination is obtained according to the branch connection point and the optimized position. If the voltage drop amplitude of the new position combination meets the preset conditions, the new position combination is determined as the final optimized position.

[0013] In a possible implementation manner of the first aspect, performing grid boundary correction based on an initial grid division result to obtain a final result of distribution network grid division includes: Constructing a coverage area with a preset radius according to the load center point of each network in the initial grid division result to obtain multiple coverage areas; Calculate the proportion of the intersection area of ​​each coverage area and the adjacent coverage area, and determine whether the proportion exceeds the preset proportion of the total area of ​​the coverage area. If it exceeds, calculate the voltage distribution of the coverage area. If the voltage distribution reaches the rated voltage, adjust the boundary of the grid corresponding to the coverage area according to the preset offset to obtain the adjusted grid boundary; According to the adjusted grid boundaries, the final result of the distribution network grid division is obtained.

[0014] In order to solve the same technical problem, a second aspect of an embodiment of the present invention provides a distribution network full-process control method, the method comprising: Obtaining a final result of the distribution network grid division, wherein the final result of the distribution network grid division is obtained by the distribution network grid division method based on artificial intelligence as in the first aspect; According to the final result of the distribution network grid division, the equipment operation status data is collected according to the sampling interval, and the voltage, current and temperature in the operation status data are quantitatively scored to obtain the equipment status mapping matrix; A construction constraint control matrix is ​​constructed based on the equipment status mapping matrix. Based on the control matrix, the construction process monitoring data is compared with the equipment rating to obtain the construction quality assessment index. The equipment operation status score is calculated based on the construction quality assessment index. When the equipment operation status score is lower than the preset operation status assessment threshold, the equipment is marked as a warning state, and the equipment maintenance cycle is generated based on the warning state. The equipment status changes during the equipment maintenance cycle are quantified and statistically analyzed to obtain statistical results. The maintenance quality score is calculated based on the statistical results. The maintenance plan is revised based on the maintenance quality score to obtain an updated equipment inspection and maintenance plan.

[0015] The technical solution of the present invention has the following advantages: The distribution network grid division method based on artificial intelligence provided by the embodiment of the present invention obtains the terrain characteristic parameters and the power load distribution of the first area in the target distribution network grid division area, performs quantitative analysis on the terrain characteristic parameters, obtains the winding degree index of the first area, obtains the terrain characteristics according to the winding degree index, and obtains the cable channel path according to the minimum curvature principle and the terrain characteristics; determines multiple grid nodes according to the power load distribution, optimizes each grid node, obtains multiple substation locations to be laid out, determines the target substation according to each substation location, obtains the optimal path of the cable channel path according to the target substation, corrects the optimal path according to the winding degree index, and obtains the actual power supply range boundary of the target substation; obtains the cable segmentation area based on the cable load value and the winding degree index in the actual power supply range boundary, uses the dynamic programming algorithm to calculate the cable segmentation area, obtains multiple alternative joint positions of cables, and screens the alternative joint positions of each cable according to the temperature rise curve along the cable to obtain Multiple intermediate joint installation positions of cables within the actual power supply range; use the analytic hierarchy process to calculate the annual maintenance cost, terrain restriction data and construction cost data of each intermediate joint installation position, obtain the evaluation score of each intermediate joint installation position, calculate the distance from each intermediate joint installation position to the actual power supply boundary, obtain multiple distance values, and determine the intermediate joint installation position with a distance value less than the preset distance value and an evaluation score less than the preset evaluation value as the position to be optimized; establish a polar coordinate sampling network according to the position to be optimized, determine the sampling point position in the polar coordinate sampling network, calculate the power supply distance and voltage drop amplitude of each sampling point position, select the sampling point position whose voltage drop amplitude meets the preset conditions as the optimized position, and obtain the final optimized position according to the load density distribution within the actual power supply range boundary; divide the power supply area of ​​the distribution network according to the actual power supply range boundary and the final optimized position, obtain the initial grid division result, perform grid boundary correction based on the initial grid division result, and obtain the final result of the distribution network grid division. Through the above division method, the reliability of the distribution network grid division result is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 A grid partitioning flow chart of a distribution network grid partitioning method based on artificial intelligence in an embodiment of the present invention; Figure 2It is a flow chart of the control method of the full-process control method of the distribution network in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The distribution network grid division method based on artificial intelligence provided by the embodiment of the present invention is as follows: Figure 1 As shown, Figure 1 The flowchart of the distribution network grid division method based on artificial intelligence includes steps S101 to S106, and each step is as follows: S101. Obtain terrain characteristic parameters and power load distribution of a first area within the target distribution network grid division area, perform quantitative analysis on the terrain characteristic parameters, obtain a winding degree index of the first area, obtain terrain characteristics based on the winding degree index, and obtain a cable channel path based on the minimum curvature principle and the terrain characteristics.

[0020] In this embodiment, the terrain elevation data of the grid area is read from the geographic information database, the terrain relief value is obtained according to the standard deviation of the elevation within the unit area, the terrain point cloud data is collected by a three-dimensional laser scanner, and the slope classification data is obtained according to the height difference ratio of adjacent sampling points, wherein the height difference ratio is divided into flat land between 0 and 15 degrees, gentle slopes between 15 and 30 degrees, and steep slopes above 30 degrees. The cross-section data is collected according to the density distribution of contour lines, and the regional tortuosity index is obtained by the ratio of the actual length of the cross-section curve to the straight-line distance between the starting and ending points. The slope classification data is superimposed on the surface elevation grid, and the terrain features are extracted from the terrain relief value and the tortuosity index using a support vector machine, and the cable channel path is obtained according to the minimum curvature principle.

[0021] In one embodiment, the terrain characteristic parameters are quantitatively analyzed to obtain the meandering index of the cable channel in the first area, including: According to the terrain characteristic parameters, the height difference ratio between adjacent sampling points in the first area is calculated, and the slope of the sampling point is obtained according to the height difference ratio; Calculate the contour line density distribution in the first area, sample the terrain in the first area at preset intervals based on the contour line density distribution, obtain cross-sectional data, and obtain a winding degree index based on the cross-sectional data and the straight-line distance corresponding to the cross-sectional data.

[0022] In this embodiment, the terrain elevation data of the grid-divided area is read from the geographic information database, the elevation standard deviation within the unit area is calculated to obtain the terrain undulation value, and the surface slope value is obtained by dividing the elevation difference of adjacent grid points by the horizontal distance. 100 terrain point cloud data are collected per square meter by a three-dimensional laser scanner, and the height difference ratio of adjacent sampling points is calculated based on the point cloud data. The slope range is divided into flat land in the range of 0 to 15 degrees, gentle slope in the range of 15 to 30 degrees, and steep slope in the range of 30 degrees. The density distribution of contour lines is calculated according to the length of contour lines within the unit area, and the terrain is sampled at a section interval of 20 meters to obtain the longitudinal section curve. The regional sinuosity index is calculated based on the ratio of the actual length of the section curve to the straight-line distance between the starting and ending points.

[0023] It should be noted that the terrain feature parameters refer to terrain point cloud data.

[0024] Then, the slope classification data is superimposed on the surface elevation grid, and the support vector machine is used to extract terrain features from the terrain undulation value and the winding index, and the path direction of the cable trench in each slope interval is calculated according to the minimum curvature principle. The terrain elevation data in the geographic information database contains the coordinates of the elevation point location and the corresponding elevation value. The grid vertex data is selected from it to form a regular grid. The height difference between two adjacent elevation points divided by the horizontal distance is the slope of the area. The standard deviation of the elevation point height reflects the degree of terrain undulation. The larger the standard deviation, the greater the terrain elevation undulation. In the mountainous scene, the standard deviation of the 12 elevation points is 52 meters, reflecting that the terrain in this area is relatively undulating. In the hilly area, the standard deviation of the 12 elevation points is 23 meters, which belongs to medium undulating terrain. From the numerical value, it can be seen that the difference between mountainous and hilly terrain is obvious.

[0025] When three-dimensional laser scanning is used to collect terrain point cloud data, 100 points are collected per square meter to ensure data density. The point cloud data obtained by scanning contains spatial coordinate information. The local slope is obtained by calculating the elevation difference between adjacent points and dividing it by the horizontal distance. When the slope value is within the range of 0 to 15 degrees, it is judged to be flat land, which is suitable for laying cables in a straight line. A slope of 15 to 30 degrees is a gentle slope, which needs to be laid along the contour line. A slope greater than 30 degrees is a steep slope, which needs to be avoided or special engineering measures should be taken. A cross-section curve is sampled every 20 meters along the terrain direction, and the ratio of the actual cross-section curve length to the straight-line distance between the start and end points is calculated to obtain the winding index.

[0026] If the meandering index is greater than 1.3, it means that the terrain is undulating and the cable must be laid around steep slopes. For areas with a meandering index less than 1.1, the terrain is gentle and suitable for straight-line cable laying. The more mountainous the terrain, the greater the meandering index, and the more the cable trench must avoid steep terrain areas. The support vector machine extracts features from the terrain undulation value and the meandering index, combines the slope classification data, and plans the cable trench path according to the minimum curvature principle. In mountainous scenarios, areas with large terrain undulations usually avoid steep slopes and detour along gentle slopes. In gentle areas, the cable trench is laid in a straight line as much as possible to save engineering work. The path calculated by the minimum curvature principle satisfies the terrain constraints and minimizes turns. Arc transitions are used at corners to avoid too small a turning radius. According to actual cases, a 500-meter-long cable trench in a mountainous area was actually laid to a length of 680 meters after optimization, with an additional detour distance of 180 meters to avoid steep slopes.

[0027] The tortuosity index reflects the degree of deviation between the actual path and the straight-line distance. In plain areas, the path is basically laid along a straight line, and the tortuosity index is close to 1.1. In hilly areas, the tortuosity index increases to about 1.3 due to the need to bypass areas with larger slopes. When the path angle exceeds 45 degrees, it indicates that there are significant terrain obstacles or impassable areas. At this time, the correction factor of 1.2 should be multiplied by the path section, reflecting the increased construction difficulty and line loss at the turn. In the power supply path planning, a total of 12 trunk paths are planned within a radius of 3 kilometers of a substation. The total impedance value of the path is obtained by calculating the product of the unit length impedance of each path and the actual length. The path combination with the minimum total impedance includes 4 trunk paths, which extend in the four directions of southeast, northwest, and northeast, respectively, forming the actual power supply range of the substation. On the eastward path, due to passing through a mountain with a slope of 25 degrees, the actual length of the path reaches 2.4 kilometers, which is 0.6 kilometers longer than the straight-line distance, but still better than other optional paths. Therefore, the end point of this path is the power supply boundary point in this direction, marking the effective power supply range of the substation in this direction.

[0028] When calculating the cable trench winding index, the elevation point data of the cable trench area is extracted from the geographic database at a sampling interval of 20 meters, and 400 sampling points are obtained within each square kilometer. The height difference between adjacent elevation points is calculated and divided by the horizontal distance to obtain the slope value, which is divided into 5 levels within the range of 0 to 45 degrees. The standard deviation of the elevation point is calculated by the squared deviation formula, which is: in, represents the standard deviation of the sample elevation, represents the number of samples, Indicates The height of the samples, Represents the average value of the sample elevations.

[0029] The relative height difference is obtained by subtracting the regional average elevation from each sampling point, and the terrain relief is calculated from the maximum height difference between adjacent sampling points. The slope value is multiplied by a weight coefficient of 0.4, and the terrain relief is multiplied by a weight coefficient of 0.6. The comprehensive index value is obtained by weighted summation, and the quantitative result between 0 and 1 is obtained by normalization based on the maximum value of the index. The cable channel meandering index is calculated based on the normalized results of the slope classification value and the terrain relief, and the weighted average of the slope weight and the relief weight is linearly mapped to the preset meandering index range. The sampling of elevation points in the geographic database is crucial to accurately reflect the terrain characteristics. At a sampling interval of 20 meters, 400 elevation points are obtained per square kilometer area, forming a 20x20 regular grid.

[0030] As an example of this embodiment, taking a mountainous area as an example, the elevation values ​​of 400 sampling points range from 320 meters to 580 meters, and the maximum height difference of adjacent points reaches 26 meters, reflecting the drastic changes in the terrain of the area. By calculating the ratio of the height difference and horizontal distance of adjacent elevation points, the slope data range is between 5 degrees and 42 degrees, and is divided into 5 levels according to the slope value: 0 to 9 degrees is a gentle area, 9 to 18 degrees is a gentle slope area, 18 to 27 degrees is a medium slope area, 27 to 36 degrees is a steep slope area, and 36 degrees or more is a steep slope area. The standard deviation calculation reflects the degree of discreteness of the terrain. In the same mountainous area, the average elevation of 400 sampling points is 450 meters, and the standard deviation value of 38 meters is calculated by the sum of squares of the deviations, indicating that the terrain in this area is relatively undulating. The relative height difference reflects the changing characteristics of the local terrain. The elevation value of a certain sampling point is 15 meters higher than the average of the surrounding 8 adjacent points, indicating that this place is a protruding part of the local terrain. The maximum height difference between adjacent sampling points in the area reaches 26 meters, and the corresponding terrain undulation is relatively high. In the weight distribution, the terrain undulation accounts for a weight of 0.6, and the slope value accounts for a weight of 0.4. This distribution reflects the dominant influence of terrain undulation on the degree of cable trench winding. Taking a certain area as an example, the terrain undulation is 0.85, the slope classification value is 0.65, and the comprehensive index value is 0.77 after weighted calculation. Dividing this value by the maximum comprehensive index value of 0.92 in the area, the normalized quantitative result of 0.84 is obtained. When calculating the cable trench winding index based on the normalized result, a linear mapping method is used. The normalized value of a certain area is 0.84, and the corresponding winding index is 1.45, indicating that the actual laying length of the cable trench in this area is 45% longer than the straight-line distance. This increase is mainly due to bypassing areas with large height differences. In actual projects, when encountering areas with a standard deviation of more than 35 meters, a detour solution is often adopted. In flat areas with a standard deviation of less than 15 meters, the cable trench is basically laid along a straight line. In this way, unfavorable terrain is avoided and excessive increase in line length is controlled.

[0031] S102. Determine multiple grid nodes according to the distribution of power load, optimize each grid node, obtain multiple substation locations to be laid out, determine the target substation according to the locations of each substation, obtain the optimal path of the cable trench path according to the target substation, correct the optimal path according to the winding degree index, and obtain the actual power supply range boundary of the target substation.

[0032] In this embodiment, the power consumption per unit area of ​​the grid area in the first area is then obtained, the load density index is calculated according to the power consumption data and divided by the maximum load density value of the area to obtain a normalized load density distribution curve, and the load center coordinate point is determined by the distribution curve. A layout constraint boundary is constructed according to the load center coordinate point and the power supply point position, grid nodes are set within the boundary range, and the grid nodes are pre-screened based on the power supply radius to obtain the screened grid nodes.

[0033] Then, the simulated annealing algorithm is used to encode the screened grid nodes, and the target fitness value is calculated by multiplying the power supply distance by the weight plus the load density by the weight; if the change rate of the target fitness value of adjacent iterations is less than the preset threshold, the location of the substation to be laid out is determined from the corresponding grid node with the smallest fitness value after iterative optimization.

[0034] Then, the target substation is determined in the substation location to be laid out, and the terrain slope data of the area around the target substation in the geographic information database is obtained. The impedance distribution along the search grid node is calculated based on the terrain slope data and the benchmark impedance value. According to the impedance distribution along the search grid node, the depth-first algorithm is used to select the adjacent grid points with the smallest impedance value for path traversal to obtain the traversal path set. For the traversal path set, the path winding index is calculated by the ratio of the actual length of the path to the straight-line distance, and the paths with the path winding index within the preset range are screened. According to the filtered path, the coordinate point sequence is extracted to construct the power supply path combination, and the total path impedance value is calculated by multiplying the unit length impedance by the path length, and the boundary point of the power supply range is determined according to the total path impedance value.

[0035] In one embodiment, a plurality of grid nodes are determined according to the distribution of power load, and each grid node is optimized to obtain the location of the substation to be laid out, including: A load density index is calculated according to the distribution of power load, and a normalized load density distribution curve is obtained according to the load density index and a load density value greater than a preset density value in the first area; Determine the load center coordinate point according to the normalized load density distribution curve, construct a boundary range according to the load centerline coordinate point and the power point position in the first area, and set grid nodes at a preset interval within the boundary range to obtain a plurality of grid nodes; Each grid node is screened according to a preset power supply radius to obtain a plurality of screened grid nodes, and a simulated annealing algorithm is used to optimize each screened grid node to obtain a plurality of substation locations to be laid out.

[0036] In this embodiment, the power consumption per unit area in the grid area is collected, the load density index is calculated based on the power consumption data, and then the load density index is normalized by dividing it by the maximum load density value of the area, and the load center coordinates of each area are calculated according to the normalized load density distribution curve. The layout constraint boundary is constructed according to the power point location and the load center coordinates, and uniform grid nodes are set at 200-meter intervals within the boundary range. The grid nodes are pre-screened based on the power supply radius of 1 to 3 kilometers to obtain the pre-screened grid nodes.

[0037] The power supply coverage area of ​​each node is calculated from 315 kVA to 1000 kVA distribution transformer capacity specifications. The simulated annealing algorithm is used, the initial temperature value is set to 100, the temperature is decreased at a rate of 0.95, the pre-screened grid nodes are encoded, the power supply distance is multiplied by 0.6 and the load density is multiplied by 0.4 to calculate the target fitness value, and the target fitness value is calculated repeatedly within the preset number of iterations. The target fitness values ​​of 50 consecutive iterations are compared. When the change rate is less than 0.1%, it is determined that the optimization convergence condition is met. From the corresponding grid node with the smallest fitness value after iterative optimization, the substation layout coordinate points are determined based on the power consumption distribution per unit area and the power supply radius constraint.

[0038] As an example of this embodiment, the distribution network monitoring device collects the power load data in the area every 5 minutes, obtains 288 sampling point data within 24 hours, and statistically obtains the daily maximum load value per unit area. When the area of ​​a residential area is 1 square kilometer, the daily maximum load density reaches 6,000 kilowatts, the daily maximum load density of the commercial area with the same area reaches 8,000 kilowatts, and the industrial area reaches 12,000 kilowatts. By dividing by the regional maximum load value of 12,000 kilowatts, the normalized load densities are 0.5, 0.67 and 1 respectively. The load density value at the load center is significantly higher than that in the surrounding area. The layout constraint boundary is determined based on the location characteristics of the power point and the load center. The line connecting the power point and the load center is used as the main axis, and the boundary is delineated within 3 kilometers on both sides of the axis. Grid nodes are evenly arranged at intervals of 200 meters within the boundary to form potential substation locations. Considering four distribution transformer capacity specifications of 315 kVA, 500 kVA, 800 kVA, and 1000 kVA, the power supply radius from the grid node to the power load point varies from 1 to 3 kilometers. The 1000 kVA distribution transformer is used in the industrial load area, the 800 kVA distribution transformer is used in the commercial area, and the 500 kVA distribution transformer is used in the residential area. The simulated annealing algorithm starts from a temperature value of 100, and the temperature is multiplied by a decay coefficient of 0.95 for each iteration. In the iterative optimization process, the power supply distance and load density are two mutually restrictive factors. The shorter the power supply distance, the smaller the line loss but the less load is covered; the more concentrated the load density, the greater the covered load but the power supply distance may increase. The two factors are balanced by weighted coefficients of 0.6 and 0.4 to calculate the target fitness value. Iterate 100 times at each temperature and record the current optimal solution. When judging the convergence of the algorithm, compare the changes in the target fitness values ​​of the last 50 iterations. If the change rate is continuously less than 0.1%, it means that a relatively stable local optimal solution has been found.

[0039] As an example in this embodiment, the target fitness value is 0.85 in the initial stage, and converges to 0.42 after 2000 iterations. The final substation location is 1.2 kilometers away from the load center and is located between two major power consumption areas. This location not only ensures a reasonable power supply radius, but also achieves effective coverage of high load density areas. 98% of the power load points within the power supply range are within 2 kilometers of the substation, meeting the power supply radius requirements.

[0040] In one embodiment, according to the target substation, the optimal path of the cable channel path is obtained, and the optimal path is corrected according to the winding degree index to obtain the actual power supply range boundary of the target substation, including: Calculate the impedance value along the grid node to be searched according to the terrain slope data and the reference impedance value of the target substation; A depth-first algorithm is used to traverse the paths of the grid nodes whose impedance values ​​along the path are less than a preset impedance value threshold, and a traversal path set is obtained; The actual length of each path in the traversed path set is calculated with the straight-line distance corresponding to each path to obtain the meandering index of each path, and the coordinate point sequence of the path whose meandering index is less than the preset meandering index is constructed to obtain multiple power supply path combinations; Based on the impedance value and the length of each path, the total impedance value of each power supply path combination is obtained, and the power supply path combination with a total impedance value less than a preset total impedance value threshold is selected as the actual power supply range boundary.

[0041] In this embodiment, the terrain slope data of the area around the target substation in the geographic information database is obtained, and the impedance distribution along the search grid node is calculated according to the terrain slope data and the benchmark impedance value; according to the impedance distribution along the search grid node, the adjacent grid points with the smallest impedance value are selected by the depth-first algorithm for path traversal to obtain a traversal path set; for the traversal path set, the path winding index is calculated by the ratio of the actual length of the path to the straight-line distance, and the paths with the path winding index within a preset range are screened; according to the screened paths, the coordinate point sequence is extracted to construct a power supply path combination, the total path impedance value is calculated by multiplying the unit length impedance by the path length, and the power supply range boundary point is determined according to the total path impedance value.

[0042] Specifically, after obtaining the terrain slope data of the area around the target substation, the impedance distribution along the route is calculated according to the line benchmark impedance value of 0.2 ohms per kilometer, the search grid nodes are divided into 50-meter intervals, and the obstacle weight value of 999 is marked for the impassable area. Then the depth-first algorithm is used to traverse the path in the search grid, and the adjacent grid point with the smallest impedance value is selected as the next search node each time. The path cost is calculated by accumulating the path segment impedance value and the obstacle weight, and the traversed path is sorted according to the cost value. The winding index is calculated according to the ratio of the actual length of the path to the straight-line distance, and the paths with the winding index in the range of 1.1 to 1.5 are screened. The turning point of the path is determined based on the 45-degree angle threshold, and the path at the turning point is multiplied by a correction coefficient of 1.2 to recalculate the path impedance. The coordinate point sequence is extracted from the screened path to construct a power supply path combination, and the total path impedance is calculated by multiplying the unit length impedance by the path length. The power supply radius is calculated cumulatively along the path, and the actual power supply range boundary point is determined on the path combination with the smallest total impedance.

[0043] The terrain slope data in the geographic information database reflects the terrain characteristics of different regions. In a certain mountainous area, the slope values ​​range from 10 degrees to 35 degrees, which causes the cable laying line to detour. The search space is divided into 50-meter grid intervals, forming 4,000 grid nodes in an area of ​​10 square kilometers. The line baseline impedance is 0.2 ohms per kilometer. In areas where the slope exceeds 30 degrees, the actual line impedance will increase to 0.4 ohms per kilometer. For impassable areas such as buildings and water systems, an obstacle weight value of 999 is set so that the search algorithm automatically avoids these areas. Then, a depth-first search is performed starting from the substation location, and each time the point with the smallest impedance is selected from the adjacent 8 grid points as the next search location. Specifically, from the substation to a power load point within a power supply radius of 2 kilometers, the algorithm searches more than 200 grid nodes, and each grid node records the cumulative path cost from the starting point to the point, including the superposition of the impedance value along the way and the obstacle weight. In addition, multiple feasible paths are found during the search process and are sorted from small to large according to the cost value. The path with the smallest cost value is often the optimal path after bypassing the obstacle.

[0044] S103. Based on the cable load value and the winding degree index in the actual power supply range boundary, the cable segmentation area is obtained, and the cable segmentation area is calculated using a dynamic programming algorithm to obtain alternative joint positions of multiple cables. According to the temperature rise curve along the cable, the alternative joint positions of each cable are screened to obtain multiple intermediate joint installation positions of the cables within the actual power supply range.

[0045] In this embodiment, the current load value within the actual power supply range is obtained, the voltage drop is calculated according to the current load value and the path length value, and the cable segmentation area is obtained through the voltage drop; the cable segmentation area is processed by a dynamic programming algorithm, the alternative joint coordinate points are calculated according to the dynamic programming algorithm, and the conductor temperature data is collected from the alternative joint coordinate points; the maximum shear stress distribution value is calculated for the alternative joint coordinate points, the stress limitation area is determined according to the maximum shear stress distribution value, and the alternative joint coordinate points are screened through the stress limitation area; the temperature rise curve along the cable is generated according to the alternative joint coordinate points, and if the temperature rise curve is lower than the temperature threshold and the maximum shear stress distribution value is lower than the material strength limit, the alternative joint coordinate point is determined to be the intermediate joint installation point.

[0046] In one embodiment, based on the cable load value and the winding degree index in the actual power supply range boundary, the cable segment area is obtained, including: The current load value within the boundary of the actual power supply range is obtained, and the voltage drop is obtained according to the current load value and the winding degree index. If the voltage drop is greater than the preset drop threshold, the path within the actual power supply range is segmented to obtain the cable segment area.

[0047] In this embodiment, the current load value within the actual power supply range boundary is obtained, and the voltage drop is calculated based on the path length value and the winding degree index. When the voltage drop exceeds 5%, the laying path is segmented and the appropriate specification is selected within the cable cross-section range of 240 to 400 square millimeters.

[0048] In one embodiment, a dynamic programming algorithm is used to calculate the cable segment area to obtain multiple candidate joint positions. The candidate joint positions of each cable are screened according to the temperature rise curve along the cable to obtain multiple intermediate joint installation positions of the cable within the actual power supply range, including: According to the cable segment area and the set search depth threshold, a dynamic programming algorithm is used to determine multiple candidate joint positions within a preset spacing range; Collect conductor temperature data at each candidate joint location; Perform stress analysis on each candidate joint position to obtain the maximum shear stress distribution value of each candidate joint position, determine the stress limit area according to the maximum shear stress distribution value, and select the candidate joint position to be arranged through the stress limit area; If the conductor temperature data corresponding to the alternative joint position to be arranged is less than the preset temperature threshold, and the maximum shear stress distribution value corresponding to the alternative joint position to be arranged is less than the preset material strength, the alternative joint position to be arranged is determined to be the intermediate joint installation position.

[0049] In this embodiment, a dynamic programming algorithm is used to calculate the cable segmentation scheme, the maximum search depth is set to 100 layers, the candidate joint positions are marked within the interval range of 500 to 800 meters, the conductor temperature data is collected from each candidate position, and the temperature distribution curve along the cable is generated. The maximum shear stress criterion is used to calculate the stress distribution value at the cable bend, the stress concentration area is marked as the joint arrangement restriction point, the candidate joint positions are screened according to the restriction point position, and the joint installation area is set where the stress value is less than the material strength limit. The temperature rise curve along the cable is generated based on the current load distribution, and the section with a temperature below 90 degrees is identified from the temperature rise curve. The dynamic programming result is superimposed on the temperature rise limit section, and the intermediate joint coordinate point is determined in the section that meets the stress limit and the temperature rise limit.

[0050] The current load value directly determines the voltage drop and temperature rise of the cable. In a 10 kV distribution line, the load current gradually decreases from 280 amperes at the outlet of the substation to 120 amperes at the end. When the total length of the cable reaches 1200 meters and the winding index is 1.3, the calculated voltage drop amplitude is 5.8%, which exceeds the limit of 5%. At this time, it is necessary to divide the line into two sections and select a cable with a cross-section of 300 square millimeters to control the voltage drop in each section within 3%. The dynamic programming algorithm sets a calculation node every 50 meters along the cable at a search depth of 100 layers. A total of 24 nodes are set on the 1200-meter line, and each node records the cumulative voltage drop and temperature rise from the starting point to that point. At a certain node, the temperature data shows that the conductor temperature reaches 85 degrees, close to the limit of 90 degrees, indicating that it is not suitable to set a joint at this point. Through dynamic programming calculations, two alternative joint locations that meet the installation spacing requirements are found at 600 meters and 750 meters respectively. In the stress analysis of the cable bends, the bend angle reached 45 degrees at one point, and the calculated maximum shear stress reached 75% of the material strength. The 100-meter range before and after this bend point was marked as a restricted area for joint arrangement. At another 30-degree bend, the stress value was only 45% of the material strength, and joints could be considered at this location. By superimposing and analyzing the stress distribution of the entire cable, five stress concentration areas were identified, and these areas were excluded from the joint installation plan.

[0051] In the complete temperature rise curve, the temperature at the outgoing end of the substation reaches 88 degrees due to the maximum current, the temperature in the middle section fluctuates between 75 and 82 degrees, and the temperature at the end drops to 65 degrees. Combining the optimization results of dynamic programming and stress constraints, the final position of the middle joint was determined 720 meters away from the substation. The conductor temperature at this position is 78 degrees, which is lower than the 90-degree limit; the local stress is only 38% of the material strength, with sufficient mechanical strength margin; and it is located in a relatively flat section between two stress concentration areas. Through this multi-dimensional comprehensive optimization, both the electrical performance of the cable operation and the mechanical reliability of the joint installation position are guaranteed. At the same time, the selection of this position also balances the voltage drop distribution of the front and rear sections of the cable, with a voltage drop of 2.8% in the front section and 2.6% in the rear section, which is basically balanced.

[0052] The winding degree index in the cable channel path is identified. If the winding degree is less than the set winding degree index threshold, the cable channel path is considered to belong to the flat area, and the cable electrical parameters are obtained to calculate the maximum allowable distance between adjacent joints. Under the condition of meeting the maximum allowable distance, the cable intermediate joints are arranged in the flat area. If the distance between two flat areas is greater than the preset threshold, the additional intermediate joints are arranged at the lowest point of the winding degree between the two flat areas. An optimization model is established with the goal of minimizing the sum of the winding degree indexes, and the position of the cable intermediate joints is optimized to obtain the layout plan of the cable intermediate joints.

[0053] The winding degree value is obtained by calculating the ratio of the actual length between adjacent points to the straight-line distance according to the coordinates of the path points, and the range of the flat area is determined from the winding degree value; the interval distance between adjacent flat areas is calculated according to the range of the flat area, and the segment separation position is determined by the interval distance; the genetic algorithm is used to generate the initial joint layout plan within the flat area, and the supplementary layout position is obtained from the local minimum winding point; the sum of the winding degree values ​​of each section is calculated for the initial joint layout plan, and the final intermediate joint layout coordinates are obtained through iterative optimization of the genetic algorithm.

[0054] In addition, specifically, the coordinates of the cable channel path points are collected at intervals of 20 meters, the ratio of the actual length to the straight-line distance at each path point is calculated to obtain the winding degree value, the flat area range is marked from the path points with a winding degree value less than 1.2, and the maximum allowable distance between adjacent joints is obtained in the range of 500 to 800 meters. The interval distance between adjacent flat areas is calculated by the path point coordinates, the section separation position is recorded for areas with an interval greater than 1000 meters, and the local minimum winding point in the non-flat area is calculated based on the deviation rate between the actual path length and the straight-line distance.

[0055] In the formula, Indicates the deviation rate based on the actual path length and the straight-line distance, represents the actual path length, Represents the straight-line distance between the first and last points.

[0056] The genetic algorithm was used to generate the initial joint layout scheme within the marked flat area starting from 100 initial solutions. The local minimum winding point was added as a supplementary layout position for the sections that did not meet the maximum allowable distance limit. The path was segmented according to all joint layout positions, and the sum of the winding degree values ​​of each section was calculated. The genetic algorithm was set with an upper limit of 200 iterations. The joint layout position was adjusted in each iteration, and the final layout coordinates were obtained based on the criterion of minimizing the sum of the winding degree. The winding degree of the cable channel path is measured by the ratio of the actual length to the straight-line distance. With a sampling interval of 20 meters, 100 path point coordinates were obtained on a 2-kilometer-long channel. The actual length of a certain section of the path was 560 meters, and the straight-line distance was 480 meters. The calculated winding degree value was 1.17, which was less than the threshold of 1.2, and the section was determined to belong to the flat area. In the flat area, the maximum allowable distance between adjacent joints depends on the cable specifications. For a cable of 300 square millimeters, the maximum allowable distance is set to 750 meters. The distribution of flat areas is often discontinuous. In a channel, three flat areas are identified, the first one is 420 meters long, the second one is 380 meters long, and the third one is 460 meters long. The interval between the first and second one is 1200 meters, which exceeds the threshold of 1000 meters and requires a joint in the middle. By calculating the winding degree of each path point in this 1200-meter section, a local minimum point with a winding degree of 1.25 is found as a potential joint layout location. The genetic algorithm starts with 100 random layout plans. In the initial plan, the probability of the joint being arranged in the flat area is high. For each layout plan, check whether the distance between adjacent joints meets the 750-meter limit. If a certain distance exceeds the limit, the local minimum winding point identified previously is introduced as the new joint location in this section. In one iteration, a layout plan contains 5 joints, 3 of which are located in the flat area and 2 are located at the local minimum winding point. The algorithm continuously optimizes the joint location after 200 iterations. The sum of the winding degree of the entire path is calculated in each iteration. In the initial stage, the total winding degree of a certain layout scheme was 6.8; after optimization, it was reduced to 5.4. The final joint layout coordinates include: 3 points in the flat area, with coordinates of (350, 420), (1200, 380) and (2100, 460), and two local minimum winding points (800, 350) and (1600, 400). This layout scheme not only meets the joint spacing restrictions, but also makes the most of the flat terrain, so that the overall winding degree is minimized. The joints arranged in the flat area are convenient for construction and maintenance, and the joints added at the local minimum winding points solve the problem of excessive spacing between adjacent flat areas.

[0057] S105. Calculate the annual maintenance cost, terrain restriction data and construction cost data of each intermediate joint installation location using the analytic hierarchy process to obtain an evaluation score for each intermediate joint installation location, calculate the distance from each intermediate joint installation location to the actual power supply boundary, obtain multiple distance values, and determine the intermediate joint installation location with a distance value less than a preset distance value and an evaluation score less than a preset evaluation value as a location to be optimized.

[0058] In this embodiment, the fault records and maintenance records of the intermediate joint installation position are obtained, and the annual maintenance cost value is calculated based on the fault records and maintenance records; a third-order judgment matrix is ​​established based on the annual maintenance cost value, terrain restriction data, and construction cost data, and the judgment matrix is ​​evaluated by the scale value to obtain a eigenvector; the eigenvector is normalized to obtain a weight vector, and if the consistency ratio of the weight vector is less than a preset threshold, the terrain restriction weight value, the maintenance cost weight value, and the construction cost weight value are extracted from the weight vector; a weighted calculation is performed on the three indicator data and the weight value of the intermediate joint installation position to obtain a comprehensive score, the comprehensive score is interval mapped to obtain an evaluation score, and a ranking result of the intermediate joint installation position is generated according to the evaluation score.

[0059] The distance value from the middle joint to the boundary is extracted according to the actual power supply range boundary, and the position to be optimized is determined for the cable joints whose evaluation scores are lower than the preset threshold and whose distance from the boundary is less than the preset range.

[0060] In one embodiment, the annual maintenance cost, terrain restriction data and construction cost data of each intermediate joint installation location are calculated using the hierarchical analysis method to obtain the evaluation score of each intermediate joint installation location, including: A judgment matrix for each intermediate joint installation location is established based on the annual maintenance cost value, terrain restriction data, and construction cost data of each intermediate joint installation location; Based on the judgment matrix of each intermediate joint installation position, the scale value is used for calculation to obtain the characteristic vector of each intermediate joint installation position, and the characteristic vector is normalized to obtain the weight vector. If the consistency ratio of the weight vector is less than the preset ratio, the terrain restriction weight value, maintenance cost weight value, and construction cost weight value are extracted from the weight vector. The terrain restriction weight value, maintenance cost weight value and construction cost weight value are weighted together with the corresponding annual maintenance cost value, terrain restriction data and construction cost data to obtain the evaluation score of each intermediate joint installation position.

[0061] In this embodiment, the fault records and maintenance records of each intermediate joint installation position in the past three years are obtained, the annual maintenance cost is calculated according to the labor hour unit price of 200 yuan per hour, and the annual failure rate index value is obtained by normalization according to the maximum value of the total number of faults at each candidate position. A three-order judgment matrix is ​​established by using hierarchical analysis, and the relative importance of the three indicators of terrain restriction, maintenance cost, and construction cost is evaluated by 1 to 9 scales. The matrix characteristic root and characteristic vector are calculated to obtain the weight vector, and the weight consistency ratio is checked to see whether it is less than 0.1. The matrix characteristic vector is normalized, and the normalized weight value is retained to 3 decimal places. The terrain restriction weight 0.3, the maintenance cost weight 0.5, and the construction cost weight 0.2 are extracted from it as the index weight value. Based on the index weight value, the three indicators of terrain restriction, maintenance cost, and construction cost of each intermediate joint installation position are weighted and summed, and the weighted result is mapped to the interval of 0 to 100 to obtain a comprehensive evaluation score, and the intermediate joint installation position is sorted in descending order according to the score value. The fault records and maintenance records of the cable joint reflect the operating conditions of different layout positions.

[0062] As an example of this embodiment, taking 5 intermediate joint installation positions as an example, the fault records in the past 3 years are 3 times, 2 times, 4 times, 1 time and 2 times, respectively, with a maximum value of 4 times. The annual failure rates obtained by normalization are 0.75, 0.5, 1.0, 0.25 and 0.5 respectively. The maintenance time record shows that each fault takes an average of 16 hours to handle. According to the labor unit price of 200 yuan per hour, the annual maintenance costs are 9600 yuan, 6400 yuan, 12800 yuan, 3200 yuan and 6400 yuan respectively. The hierarchical analysis method quantifies the relative importance between indicators by constructing a judgment matrix. In the third-order judgment matrix, the importance ratio of maintenance cost to terrain restriction is 3, and the importance ratio to construction cost is 2, indicating that maintenance cost is the most important of the three indicators. The importance ratio of terrain restriction to construction cost is 1.5, which reflects that the influence of terrain factors is second. By calculating the eigenvector of the matrix and normalizing it, the maintenance cost weight is 0.5, the terrain restriction weight is 0.3, and the construction cost weight is 0.2. The calculated value of the consistency ratio is 0.05, which is less than the threshold of 0.1, indicating that the weight distribution is reasonable. After the weight is determined, the three indicators of each intermediate joint installation location are quantified. The indicator values ​​of a certain intermediate joint installation location are: terrain restriction 0.8 (1 is the worst and 0 is the best), maintenance cost 0.6, and construction cost 0.4. The three indicator values ​​are multiplied by the corresponding weights 0.3, 0.5, and 0.2 respectively, and the sum is obtained to obtain a weighted result of 0.64. In order to make the evaluation results more intuitive, the weighted result is multiplied by 100 and mapped to the interval of 0 to 100, and the comprehensive evaluation score of the intermediate joint installation location is 36 points. This score reflects the comprehensive performance of the location in the three evaluation dimensions. The same calculation is performed on the five candidate locations, and the comprehensive evaluation scores obtained are 65 points, 78 points, 42 points, 83 points, and 71 points, respectively. Although the fourth intermediate joint installation location with the highest score has a higher construction cost, it has the best comprehensive performance due to its good terrain conditions and low maintenance cost. The third intermediate joint installation location has the lowest score, which is mainly affected by the high failure rate and maintenance cost. This multi-dimensional quantitative evaluation not only takes into account long-term operation and maintenance costs, but also takes into account terrain and construction factors, making the location selection more objective and reasonable. The use of different weights for different types of indicators also reflects the differentiated consideration of the importance of various factors in the evaluation process.

[0063] S105. Establish a polar coordinate sampling network according to the position to be optimized, determine the sampling point positions in the polar coordinate sampling network, calculate the power supply distance and voltage drop amplitude of each sampling point position, select the sampling point position whose voltage drop amplitude meets the preset conditions as the optimized position, and obtain the final optimized position according to the load density distribution within the boundary of the actual power supply range.

[0064] In this embodiment, a polar coordinate sampling grid is established according to the position to be optimized, the power supply distance and the voltage drop amplitude are calculated for the sampling grid, and new arrangement coordinate points are obtained from sampling points that meet the voltage drop amplitude threshold.

[0065] The power density value is calculated based on the load distribution map within the actual power supply range, and the cable branch power supply coverage area is calculated based on the power density value. The point with the maximum load density in the power supply coverage area is obtained as the branch connection point. The newly arranged coordinate points and branch connection points are combined in position. If the voltage drop amplitude of the combination scheme is less than the preset threshold, the coordinate combination with the highest evaluation score is determined from the combination schemes that meet the conditions.

[0066] In one embodiment, obtaining the final optimized position according to the load density distribution within the actual power supply range boundary includes: According to the load density distribution within the actual power supply range, a location where the load density is greater than a preset load density threshold is selected as a branch connection point; A new position combination is obtained according to the branch connection point and the optimized position. If the voltage drop amplitude of the new position combination meets the preset conditions, the new position combination is determined as the final optimized position.

[0067] In this embodiment, a 10-meter resolution actual power supply range boundary layer is used to extract the distance value from the middle joint position to the power supply boundary, and the joints with an evaluation score less than 60 points and less than 500 meters from the boundary are marked as points to be optimized. At the same time, the coordinate space within 200 meters around the point to be optimized is recorded, and a polar coordinate search grid is established for each point to be optimized. Equally spaced sampling points are generated within 200 meters, and the power supply distance and voltage drop amplitude at the sampling point position are calculated. New layout coordinate points are extracted from the sampling points that meet the 5% voltage drop limit.

[0068] Based on the load distribution map within the actual power supply range, the power density is calculated, and the cable branch power supply coverage area is calculated according to the 400 kVA capacity limit. The point with the most concentrated load in the power supply coverage area is selected as the branch connection point. The new layout coordinate points and branch connection points are combined, and the total voltage drop value and joint evaluation score of the combination scheme are calculated. From the scheme with a voltage drop of less than 5%, the coordinate combination with the highest evaluation score is selected as the final layout position.

[0069] It should be noted that the use of 10-meter resolution can accurately reflect the actual power supply range of the substation.

[0070] As an example of this embodiment, in a certain area, there are 8 cable joints that need to be evaluated, of which 3 joints have evaluation scores of 52 points, 48 ​​points and 55 points respectively, and the distances to the boundaries of the power supply range are 420 meters, 380 meters and 450 meters respectively. These 3 joints meet the conditions that the evaluation score is less than 60 points and the distance to the boundary is less than 500 meters, and are marked as points to be optimized. Taking one of the joints with a score of 52 as an example, a sampling grid is generated within a range of 200 meters around it. Polar coordinate search constructs sampling points in two dimensions: angle and radius. Within a range of 200 meters, a direction is taken every 15 degrees and a distance point is taken every 20 meters to form a sampling grid. For a joint to be optimized, a total of 96 sampling points are generated.

[0071] By calculating the power supply distance and voltage drop of each sampling point, it is found that the voltage drop at the sampling point 160 meters away from the original position at 30 degrees north-east is 4.2%, which meets the limit requirement of 5%. The load distribution layer shows the spatial distribution characteristics of power density. In the power supply area near the joint to be optimized, two areas with high power density are identified, with unit area load density reaching 350 kVA per square kilometer and 280 kVA per square kilometer respectively. Considering the capacity limit of 400 kVA, the maximum power supply radius of the branch cable is calculated to be 800 meters. Within this range, the center of the area with a load density of 350 kVA per square kilometer is selected as the branch connection point. The final layout plan needs to meet the requirements of voltage drop and evaluation score at the same time. After combining the new layout coordinate points with the branch connection points, multiple feasible plans are obtained. In one of the plans, the voltage drop of the trunk cable joint after moving to the new position is 4.2%, and the voltage drop of the branch cable is 3.8%, both of which meet the limit of 5%. At the same time, the evaluation score of the new location was increased to 78 points, a significant improvement over the original 52 points. This solution had the highest evaluation score among all solutions that met the voltage drop requirements, so it was selected as the final layout solution. Through this multi-dimensional optimization, the power supply quality was guaranteed and the rationality of the joint layout was improved. For the other two joints to be optimized, the same method was used to find their respective optimal layout solutions, which increased the evaluation scores from 48 points to 72 points and from 55 points to 75 points respectively.

[0072] S106. Divide the power supply area of ​​the distribution network according to the actual power supply range boundary and the final optimized position to obtain an initial grid division result, perform grid boundary correction based on the initial grid division result, and obtain a final result of the distribution network grid division.

[0073] In this embodiment, the Thiessen polygon algorithm is used to obtain the initial grid division result according to the actual power supply range boundary, the load center point coordinates are marked according to the initial grid division result, a circular coverage area is constructed for the load center point coordinates, and the cross-area ratio of adjacent areas is calculated through the circular coverage area; if the cross-area ratio exceeds a preset threshold, the grid boundary is adjusted according to the offset, and the boundary correction is performed on the area where the load center point deviation distance exceeds the set threshold; the support vector regression algorithm is used to process the load center point coordinates and the power supply radius data, the grid boundary direction is determined according to the processing results, and the grid boundary line is corrected by the grid area data and the load distribution data.

[0074] In one embodiment, a grid boundary correction is performed based on the initial grid division result to obtain a final result of the distribution network grid division, including: Constructing a coverage area with a preset radius according to the load center point of each network in the initial grid division result to obtain multiple coverage areas; Calculate the proportion of the intersection area of ​​each coverage area and the adjacent coverage area, and determine whether the proportion exceeds the preset proportion of the total area of ​​the coverage area. If it exceeds, calculate the voltage distribution of the coverage area. If the voltage distribution reaches the rated voltage, adjust the boundary of the grid corresponding to the coverage area according to the preset offset to obtain the adjusted grid boundary; According to the adjusted grid boundaries, the final result of the distribution network grid division is obtained.

[0075] In this embodiment, the Thiessen polygon algorithm is used to preliminarily divide the power supply area, and the grids are merged based on the constraint condition of the minimum area of ​​1 square kilometer, and the load center point of each grid is marked. A circular coverage area with a radius of 3 kilometers is constructed for the load center point, and the area ratio of the intersection of adjacent coverage areas is calculated. The voltage distribution in the intersection area is quantified by the voltage drop limit, and it is determined whether the intersection area exceeds 10% of the total area. In areas where the intersection area exceeds 10%, the grid boundary is adjusted according to an offset of 50 to 100 meters, and the boundary correction is performed on the area where the load center point deviates from the original position by more than 500 meters, and the boundary point is smoothed based on the 200-meter distance threshold. Support vector regression is used to establish the corresponding relationship between the load center coordinates and the power supply radius, and the final direction of the grid boundary is determined by the corresponding relationship, and the grid boundary line is corrected according to the grid area and load distribution.

[0076] As an example in this embodiment, Thiessen polygon partitioning defines the initial boundary of the power supply area based on the location of the substation. There are 5 substations in a certain area. Five initial grids are obtained by the Thiessen polygon algorithm, of which the smallest grid area is 0.8 square kilometers, which is less than the threshold of 1 square kilometer and needs to be merged with adjacent grids. The areas of the four merged grids are 2.4, 1.8, 2.2 and 1.6 square kilometers respectively. The load center in each grid is calculated by weighted average of electricity consumption, and the offset distance of the load center point relative to the substation is between 300 and 450 meters.

[0077] The circular coverage area reflects the actual power supply capacity of the substation. A circle with a radius of 3 km is drawn to form a coverage area. The coverage circles of two adjacent substations overlap, and the proportion of the overlapping area to the area of ​​a single circle is calculated. In some overlapping areas, the area ratio reaches 12%, exceeding the threshold of 10%. By calculating the voltage distribution in the overlapping area, it is found that the maximum voltage drop reaches 6.2% of the rated voltage, and the grid boundary needs to be adjusted.

[0078] The boundary adjustment gives priority to the location of the load center. For areas with excessive overlapping areas, the boundary is offset to the side with smaller load. On a certain boundary, the distance from the original boundary to the load center is 420 meters. After the 75-meter offset adjustment, the distance from the new boundary to the load center becomes 345 meters, meeting the maximum offset limit of 500 meters. After the boundary adjustment, the boundary points are screened and smoothed using a distance threshold of 200 meters, and the local jagged boundaries are deleted. Support vector regression establishes a mapping relationship between the load center location and the power supply radius. In the divided grid, when the load center is 350 meters away from the substation, the actual power supply radius is 2.8 kilometers; when the load center is 420 meters away, the power supply radius is shortened to 2.6 kilometers. This mapping relationship guides the determination of the final boundary. By fine-tuning the boundary line, the grid area is more matched with the load distribution. After adjustment, a grid has an area of ​​2.1 square kilometers, and the load center is located 380 meters northeast of the substation, with a power supply radius of 2.7 kilometers, forming an independent power supply unit with balanced load distribution and reasonable power supply range.

[0079] The whole process control method of the distribution network provided by the embodiment of the present invention is as follows: Figure 2 As shown, Figure 2 The flowchart of the whole process control method of the distribution network includes steps S201 to S204, wherein: S201. Obtain the final result of distribution network grid division.

[0080] In this embodiment, the final result of the distribution network grid division is obtained by utilizing the distribution network grid division method based on artificial intelligence in the above embodiment.

[0081] By dividing the distribution network into grids, the large and complex distribution network area can be decomposed into multiple smaller grid units with clear boundaries. Each grid is managed by a dedicated operation and maintenance personnel or team. This allows for a more detailed understanding of the equipment operation status and fault conditions in each grid, making it easier to quickly locate problems and take corresponding measures.

[0082] S202. According to the final result of the distribution network grid division, the equipment operation status data is collected at sampling intervals, and the voltage, current and temperature in the operation status data are quantitatively scored to obtain the equipment status mapping matrix.

[0083] In this embodiment, the optimization parameters of the substation layout and cable trench layout in each grid are obtained, the equipment operation status data is collected at a sampling interval of 5 minutes, the indicators such as voltage, current and temperature are quantitatively scored, and the equipment operation status mapping matrix is ​​generated from the scoring results.

[0084] S203. Construct a construction constraint control matrix according to the equipment status mapping matrix, compare the construction process monitoring data with the equipment rated value based on the control matrix to obtain the construction quality assessment index; calculate the equipment operation status score according to the construction quality assessment index, and when the equipment operation status score is lower than the preset operation status assessment threshold, mark the equipment as a warning state, and generate an equipment maintenance cycle according to the warning state.

[0085] In this embodiment, according to the equipment status mapping matrix, deep reinforcement learning is used to construct the progress quality control matrix under construction constraints, that is, the construction constraint control matrix. The construction progress deviation is controlled within 10%, the monitoring data during the construction process is compared with the equipment rating, and the construction quality evaluation index is generated from the real-time monitoring data. The equipment operation status score is calculated based on the monitoring data. When the score is lower than 80 points, it is marked as a warning state. A maintenance cycle of 3 to 6 months is automatically generated based on the equipment status score, and the maintenance time is determined by historical monitoring data.

[0086] S204, quantitatively counting the equipment status changes during the equipment maintenance cycle to obtain statistical results, calculating the maintenance quality score based on the statistical results, and revising the maintenance plan based on the maintenance quality score to obtain an updated equipment inspection and maintenance plan.

[0087] In this embodiment, the equipment status changes within the maintenance cycle are quantitatively counted, the maintenance quality score is calculated from the maintenance implementation data, the inspection frequency is corrected by less than 20% based on the score, and the equipment inspection and maintenance procedures are updated based on the correction results.

[0088] It should be noted that the maintenance plan refers to the frequency of inspections.

[0089] As an example of this embodiment, the distribution network equipment operation status data includes multiple monitoring indicators. Taking a certain cable line as an example, the operation data is collected every 5 minutes, including the current load rate, cable surface temperature, insulation resistance value, etc. When the current load rate reaches 85%, the cable surface temperature reaches 65 degrees, and the insulation resistance drops to 70% of the design value, the equipment status score drops to 75 points, triggering the warning threshold. Through 24 hours of continuous data analysis, it is found that the line is often overloaded between 14:00 and 17:00 on working days. The quality control of the construction process involves constraints in multiple dimensions. In the construction of the cable trench, the original construction period was 45 days, the actual completion time was 48 days, and the progress deviation was 6.7%, which is within the allowable range of 10%. The construction quality monitoring data shows that the minimum bending radius of the cable laying is 15 times the outer diameter of the cable, and the trench backfill density reaches 95%, and these indicators meet the design requirements. Deep reinforcement learning optimizes the construction progress while ensuring the quality of the project by setting a composite reward function of progress and quality.

[0090] The equipment operation status score directly affects the determination of the maintenance cycle. The recent status score of a certain cable section is 82 points, which is higher than the warning value of 80 points. The system automatically sets the maintenance cycle to 6 months. The maintenance cycle of another cable section, whose status score dropped to 76 points due to local overheating, was adjusted to 3 months. Historical data analysis shows that the probability of failure of equipment with status scores between 75 and 80 points within 3 months has increased significantly. Maintenance quality assessment is achieved through status comparison at multiple time points. After the maintenance of a certain warning cable, its operation status score increased from 76 points to 94 points, and the maintenance quality score was 92 points. Based on this result, the preventive maintenance cycle in the maintenance plan was shortened by 15%, and the inspection frequency was revised based on the original design parameters. The half-year tracking data showed that with the revised maintenance procedures, the status score of this type of cable remained stable at above 85 points. Through this closed-loop feedback mechanism, the operation and maintenance procedures are continuously optimized in practice, forming a data-driven intelligent operation and maintenance system. In another case, live inspection and preventive replacement of cable joints with degraded insulation performance improved the condition score of the entire line by 12 points, verifying the effectiveness of the preventive maintenance strategy.

[0091] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0092] The specific embodiments described above further illustrate the purpose, technical solutions 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. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A distribution network grid division method based on artificial intelligence, characterized in that: include: Acquire terrain characteristic parameters and power load distribution of a first area within the target distribution network grid division area, perform quantitative analysis on the terrain characteristic parameters, obtain a winding degree index of the first area, obtain terrain characteristics according to the winding degree index, and obtain a cable channel path according to the minimum curvature principle and the terrain characteristics; Determine a plurality of grid nodes according to the distribution of the power load, optimize each of the grid nodes to obtain a plurality of substation locations to be laid out, determine a target substation according to each of the substation locations, obtain an optimal path of the cable trench path according to the target substation, correct the optimal path according to the winding degree index, and obtain an actual power supply range boundary of the target substation; Based on the cable load value and the winding degree index in the boundary of the actual power supply range, a cable segmentation area is obtained, and the cable segmentation area is calculated using a dynamic programming algorithm to obtain a plurality of alternative joint positions of the cables, and the alternative joint positions of each of the cables are screened according to the temperature rise curve along the cable to obtain a plurality of intermediate joint installation positions of the cables within the actual power supply range; The annual maintenance cost, terrain restriction data and construction cost data of each intermediate joint installation position are calculated by using the hierarchical analysis method to obtain an evaluation score of each intermediate joint installation position, and the distance from each intermediate joint installation position to the actual power supply boundary is calculated to obtain multiple distance values, and the intermediate joint installation position whose distance value is less than a preset distance value and whose evaluation score is less than a preset evaluation value is determined as a position to be optimized; Establishing a polar coordinate sampling network according to the position to be optimized, determining the sampling point positions in the polar coordinate sampling network, calculating the power supply distance and voltage drop amplitude of each sampling point position, selecting the sampling point position whose voltage drop amplitude meets the preset conditions as the optimized position, and obtaining the final optimized position according to the load density distribution within the boundary of the actual power supply range; The power supply area of ​​the distribution network is divided according to the actual power supply range boundary and the final optimized position to obtain an initial grid division result, and grid boundary correction is performed based on the initial grid division result to obtain a final result of the distribution network grid division.

2. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The quantitative analysis of the terrain characteristic parameters to obtain the meandering index of the cable channel in the first area includes: Calculating the height difference ratio between adjacent sampling points in the first area according to the terrain characteristic parameters, and obtaining the slope of the sampling point according to the height difference ratio; Calculate the contour line density distribution in the first area, sample the terrain in the first area at preset intervals based on the contour line density distribution to obtain cross-sectional data, and obtain a winding degree index based on the cross-sectional data and the straight-line distance corresponding to the cross-sectional data.

3. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The step of determining a plurality of grid nodes according to the distribution of the power load, optimizing each of the grid nodes, and obtaining a substation location to be laid out includes: Calculate a load density index according to the power load distribution, and obtain a normalized load density distribution curve according to the load density index and a load density value greater than a preset density value in the first area; Determine the load center coordinate point according to the normalized load density distribution curve, construct a boundary range according to the load centerline coordinate point and the power point position in the first area, and set grid nodes at a preset interval within the boundary range to obtain a plurality of grid nodes; The grid nodes are screened according to a preset power supply radius to obtain a plurality of screened grid nodes, and the simulated annealing algorithm is used to optimize the screened grid nodes to obtain a plurality of substation locations to be laid out.

4. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The step of obtaining an optimal path of the cable channel path according to the target substation, and correcting the optimal path according to the winding degree index to obtain an actual power supply range boundary of the target substation includes: Calculate the impedance value along the grid node to be searched according to the terrain slope data and the reference impedance value of the target substation; A depth-first algorithm is used to traverse the paths of the grid nodes whose impedance values ​​along the route are less than a preset impedance value threshold, to obtain a traversal path set; The actual length of each path in the traversal path set is calculated with the straight-line distance corresponding to each path to obtain the meandering index of each path, and the coordinate point sequence of the path whose meandering index is less than the preset meandering index is constructed to obtain multiple power supply path combinations; Based on the impedance value and the length of each path, the total impedance value of each power supply path combination is obtained, and the power supply path combination whose total impedance value is less than a preset total impedance value threshold is selected as the actual power supply range boundary.

5. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The obtaining of the cable segmentation area based on the cable load value in the actual power supply range boundary and the meandering degree index comprises: The current load value within the boundary of the actual power supply range is obtained, and the voltage drop is obtained according to the current load value and the winding degree index. If the voltage drop is greater than a preset drop threshold, the path within the actual power supply range is segmented to obtain a cable segment area.

6. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The dynamic programming algorithm is used to calculate the cable segment area to obtain multiple candidate joint positions, and the candidate joint positions of each cable are screened according to the temperature rise curve along the cable to obtain multiple intermediate joint installation positions of the cable within the actual power supply range, including: According to the cable segment area, a plurality of candidate joint positions are determined by a dynamic programming algorithm within a preset spacing range according to a set search depth threshold; Collecting conductor temperature data at each of the candidate joint positions; Performing stress analysis on each of the candidate joint positions to obtain a maximum shear stress distribution value on each of the candidate joint positions, determining a stress restriction region based on the maximum shear stress distribution value, and screening out candidate joint positions to be arranged through the stress restriction region; If the conductor temperature data corresponding to the alternative joint position to be arranged is less than the preset temperature threshold, and the maximum shear stress distribution value corresponding to the alternative joint position to be arranged is less than the preset material strength, then the alternative joint position to be arranged is determined to be the intermediate joint installation position.

7. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The method of using the analytic hierarchy process to calculate the annual maintenance cost, terrain restriction data and construction cost data of each intermediate joint installation location to obtain the evaluation score of each intermediate joint installation location includes: Establishing a judgment matrix for each of the intermediate joint installation locations according to the annual maintenance cost value, terrain restriction data and construction cost data of each of the intermediate joint installation locations; Based on the judgment matrix of each of the intermediate joint installation positions, the scale value is used for calculation to obtain the characteristic vector of each of the intermediate joint installation positions, the characteristic vector is normalized to obtain the weight vector, and if the consistency ratio of the weight vector is less than the preset ratio, the terrain restriction weight value, the maintenance cost weight value, and the construction cost weight value are extracted from the weight vector; The terrain restriction weight value, the maintenance cost weight value and the construction cost weight value are weighted together with the corresponding annual maintenance cost value, the terrain restriction data and the construction cost data to obtain an evaluation score for each intermediate joint installation position.

8. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The obtaining of the final optimized position according to the load density distribution within the boundary of the actual power supply range includes: According to the load density distribution within the boundary of the actual power supply range, a position where the load density is greater than a preset load density threshold is selected as a branch connection point; A new position combination is obtained according to the branch connection point and the optimized position. If the voltage drop amplitude of the new position combination meets a preset condition, the new position combination is determined to be the final optimized position.

9. The method for gridding a distribution network based on artificial intelligence according to claim 1, characterized in that: The grid boundary correction is performed based on the initial grid division result to obtain the final result of the distribution network grid division, including: Constructing a coverage area with a preset radius according to the load center point of each network in the initial grid division result to obtain multiple coverage areas; Calculate the proportion of the intersection area of ​​each of the coverage areas and the adjacent coverage areas, determine whether the proportion exceeds the preset proportion of the total area of ​​the coverage area, and if so, calculate the voltage distribution of the coverage area, and if the voltage distribution reaches the rated voltage, adjust the boundary of the grid corresponding to the coverage area according to the preset offset to obtain an adjusted grid boundary; According to the adjusted grid boundaries, the final result of the distribution network grid division is obtained.

10. A method for controlling the entire process of a distribution network, characterized in that: include: Obtaining a final result of the distribution network grid division, wherein the final result of the distribution network grid division is obtained by the distribution network grid division method based on artificial intelligence according to any one of claims 1 to 9; According to the final result of the distribution network grid division, the equipment operation status data is collected at sampling intervals, and the voltage, current and temperature in the operation status data are quantitatively scored to obtain an equipment status mapping matrix; A construction constraint control matrix is ​​constructed according to the equipment status mapping matrix, and the construction process monitoring data is compared with the equipment rated value based on the control matrix to obtain a construction quality assessment index; the equipment operation status score is calculated according to the construction quality assessment index, and when the equipment operation status score is lower than a preset operation status assessment threshold, the equipment is marked as a warning state, and an equipment maintenance cycle is generated according to the warning state; The equipment state changes during the equipment maintenance cycle are quantitatively counted to obtain statistical results, a maintenance quality score is calculated based on the statistical results, and the maintenance plan is revised based on the maintenance quality score to obtain an updated equipment inspection and maintenance plan.

Images