Optimization method of unmanned aerial vehicle fixed nest site selection model facing power inspection requirements based on GIS-ALNS

Abstract

The invention discloses a GIS-ALNS-based optimization method of a fixed nest site selection model of an unmanned aerial vehicle for an electric power inspection demand, relates to multi-target site selection optimization, and aims to solve the problems of too high cost, relatively low efficiency and the like of manual inspection of an electric power tower. The method comprises the following steps: firstly, considering machine nest construction suitability difference in different regions, establishing a suitability grid map by using ArcGIS, and preliminarily screening out high-order points with higher suitability; secondly, comprehensively considering the tower coverage rate and the nest service performance, and designing a target function with single nest coverage overrun punishment; then, a multi-target optimization site selection model is constructed by taking the maximum coverage rate of the electric power tower and the highest average suitability of the aircrafts under the limiting conditions of the inspection radius of the unmanned aerial vehicle, the maximum number of the aircrafts and the like as targets, and finally, the multi-target optimization site selection model is solved by adopting a self-adaptive large neighborhood search algorithm. According to the method, the local optimal solution can be avoided, the coverage effect and the site selection quality are balanced, the obtained nest site selection is more scientific, the power inspection efficiency is improved, and the cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of multi-objective site selection optimization, specifically to an optimization method for a fixed UAV nesting model based on GIS-ALNS for power line inspection needs. Background Technology

[0002] With the comprehensive advancement of smart grid construction, the demand for intelligent and efficient transmission line inspection continues to grow. Traditional manual inspection methods face numerous bottlenecks, such as complex terrain, limited coverage, and low operational safety. Meanwhile, drones, with their advantages of high flexibility, high inspection efficiency, and low cost, have become a core technological solution in the power inspection field. Fixed drone nests, as key supporting facilities for autonomous drone inspection systems, directly determine the completeness of inspection coverage, operational response speed, and system operating costs through their rational site selection. They are a core prerequisite for achieving large-scale automated transmission line inspection.

[0003] The optimization of fixed UAV nesting site selection models is a composite optimization problem of set covering problem and p-median problem, a classic NP-hard problem. Most current research employs exact algorithms or traditional heuristic algorithms for solving this problem. Exact algorithms are typically only suitable for small-scale instances with small transmission line scale and simple constraints. Traditional heuristic algorithms (such as genetic algorithms and particle swarm optimization) dominate current research, but even these commonly used heuristic algorithms may take a long time to generate high-quality site selection schemes for large-scale transmission networks and are prone to getting trapped in local optima. Furthermore, in actual power line inspection scenarios, the geographical environment and inspection requirements exhibit significant spatial heterogeneity and constraint complexity, such as geographical constraints like terrain undulations, obstacle distribution, and no-fly zone designations, as well as differentiated inspection requirements for different transmission line sections, such as inspection frequency requirements and fault emergency response priorities. All of these factors affect the rationality of nesting site selection and the operational efficiency of the inspection system. However, most current studies only consider basic coverage and cost objectives, neglecting the precise characterization of geospatial constraints and the impact of differentiated inspection needs on site selection results. Furthermore, incorporating precise geospatial constraints and dynamic inspection needs will further increase the complexity of the problem, reducing the solution efficiency and reliability of traditional algorithms. Therefore, it is necessary to comprehensively consider the impact of precise geospatial constraints and differentiated inspection needs on drone nest site selection, designing an efficient optimization method based on GIS-ALNS. This method combines the spatial analysis capabilities of GIS technology with the global optimization advantages of the ALNS algorithm to accurately match the actual needs of power line inspection, quickly outputting globally near-optimal drone nest site selection schemes, and improving the operational efficiency and practicality of UAV inspection systems. Summary of the Invention

[0004] This invention discloses an optimization method for the fixed drone nesting site selection model based on GIS-ALNS for power line inspection needs, aiming to address issues such as cost and efficiency in inspecting power poles. It effectively avoids local optima, balances inspection coverage with site selection quality, achieves scientific drone nesting site selection, improves power line inspection efficiency, and reduces overall costs.

[0005] To achieve the above objectives, the technical solution provided by the present invention is as follows:

[0006] An optimization method for a fixed UAV nest location model based on GIS-ALNS for power line inspection needs includes the following steps:

[0007] Step 1: Considering the differences in the suitability of nest construction in different regions, use ArcGIS to create a suitability raster map and initially screen sites with higher suitability;

[0008] Step 2: Taking into account both tower coverage requirements and cell service performance, design an objective function with a single cell coverage over-limit penalty.

[0009] Step 3: Considering the limitations of drone inspection radius and maximum number of drone nests under real-world conditions, construct a multi-objective optimization site selection model with the objectives of maximizing power pole coverage and achieving the highest average suitability of drone nests;

[0010] Step 4: Based on the objective function, the Adaptive Large Neighborhood Search (ALNS) algorithm is used to solve the multi-objective optimal location model.

[0011] To optimize the above technical solution, the specific measures / limitations also include:

[0012] In step 1, the suitability of different regions is defined by five different attributes, including terrain features, transportation convenience, population density, land use type, and whether they are located in a no-fly zone:

[0013] The classification criteria for slope suitability are based on angle ranges, as detailed below:

[0014] 1 point (suitable height): The slope is between 0 and 5 degrees;

[0015] 3 / 4 (suitable): Slope between 5 and 10 degrees;

[0016] 2 / 4 (generally suitable): Slope between 10 and 15 degrees;

[0017] 1 / 4 (lower suitable): The slope is in the range of 15-20 degrees;

[0018] 0 points (Unsuitable): Slope is above 20 degrees.

[0019] Slope aspect suitability classification is based on azimuth range:

[0020] 1 point (suitable height): Slope facing south, corresponding to an azimuth of 90-270°;

[0021] 0 points (unsuitable): Slope facing north, corresponding to an azimuth of 0-90° or 270-360°.

[0022] Traffic suitability classification is based on the Euclidean distance range (in meters) from the road:

[0023] 1 point (suitable height): 100-500 meters from the road;

[0024] 3 / 4 (Suitable): 500-1100 meters from the road;

[0025] 2 / 4 (Generally suitable): 1100-1500 meters from the road;

[0026] 1 / 4 (lower suitable): 0-100 meters or 1500-2100 meters from the road;

[0027] 0 points (Unsuitable): More than 2100 meters away from the road.

[0028] Population density suitability classification is based on the population size per unit area (unit: people):

[0029] 1 point (Highly suitable): Population size 0-50 people;

[0030] 3 / 4 (more suitable): Population size 50-300 people;

[0031] 2 / 4 (Generally suitable): Population size 300-1000 people;

[0032] 1 / 4 (lower suitable): Population size 1000-2000 people;

[0033] 0 points (Unsuitable): Population exceeds 2,000.

[0034] The suitability classification criteria for no-fly zones are based on whether the area is located within a no-fly zone:

[0035] 1 point (suitable altitude): The area is located outside the no-fly zone;

[0036] 0 points (Unsuitable): The area is located within a no-fly zone.

[0037] The land use suitability classification standard is based on land use type:

[0038] 1 point (suitable height): Bare ground;

[0039] 3 / 4 (most suitable): Artificial surface;

[0040] 2 / 4 (generally suitable): Grassland;

[0041] 1 / 4 (lower suitability): Woodland;

[0042] 0 points (Unsuitable): Farmland, water bodies or wetlands.

[0043] The Analytic Hierarchy Process (AHP) and the entropy weighting method are used to assign weights to each attribute. The suitability of all attributes of a single grid is weighted and summed to obtain the construction suitability of the grid. The grids with the top 50% construction suitability are selected as preliminary candidate points for nest construction.

[0044] In step 2, an objective function with a single-nest coverage over-limit penalty is designed. To ensure that the number of towers covered by each nest is within its performance support range, a single-nest coverage over-limit penalty is defined: for each selected nest location, if the number of nests it covers exceeds the limit, a penalty of -1 is applied; the objective function with the single-nest coverage over-limit penalty... as follows:

[0045] in, B represents the coverage rate weight, where B is the tower coverage rate. Let A be the average nest suitability weight, γ be the average nest suitability, γ be the penalty coefficient, and P be the cumulative penalty for selecting nests that exceed the limit. The formula for calculating P is as follows:

[0046]

[0047] Where, x j Indicates whether to select candidate nest c j , This is an indicator function; if the nest is selected... If the number of covered poles exceeds the limit, return 1; otherwise, return 0.

[0048] In step 3, considering limitations such as the drone inspection radius and the maximum number of drone nests under realistic conditions, the goal is to maximize the coverage of power poles and the average suitability of drone nests, as shown in the following formula:

[0049] Max

[0050] in

[0051]

[0052] z i Indicates tower e i Whether it is covered by at least one nest, where E represents the set of poles, C represents the set of candidate nests, and sj Indicates nest c j suitability, y ij Indicates tower e i Is it nested? j Coverage, p j Indicates nest c j The penalty received for exceeding the limit.

[0053] The constraints are specifically manifested as follows:

[0054]

[0055] Where, d ij Indicates tower e i With Nest C j The Euclidean distance between them This is an indicator function, which is 1 when the condition is true and 0 otherwise. K is the maximum number of nests.

[0056] In step 4, the Adaptive Large Neighborhood Search (ALNS) algorithm is used to solve the multi-objective optimal location model. First, an initial solution is generated, and the specific steps are as follows:

[0057] Step 1: Generate initial solution using a greedy strategy: initialize the nest selection state to "not selected" and the tower coverage state to "not covered", and pre-calculate the initial coverage capability of each candidate nest;

[0058] Step 2: Iterative selection of candidate cellars: Each time, from the candidate cellars that are "not selected but still have coverage capacity", select the location with the most uncovered towers, mark it as "selected" and update the global coverage status. At the same time, dynamically reduce the remaining coverage capacity of other candidate cellars until there are no uncovered towers or the maximum number of cellars is reached.

[0059] After the initial solution is generated, the ALNS core iteration process is executed, with the following specific steps:

[0060] Step 1: Operator selection: Based on the roulette wheel method, randomly select the destroying operator and the repair operator according to the current operator weight (the higher the weight of the operator, the greater the probability of being selected), and update the operator usage count;

[0061] Step 2: Destruction Phase: Invoke the selected destruction operator to "destroy" the current solution and generate a partial solution;

[0062] Step 3: During the repair phase, the selected repair operator is invoked to "repair" the partial solutions and replenish the number of nests to the maximum limit.

[0063] After completing the core iterative process of ALNS, the solution is evaluated and accepted. The specific steps are as follows:

[0064] Step 1: Effective Coverage Calculation: For the repaired new solution, calculate the effective coverage matrix - traverse the selected cell in descending order of suitability, each cell prioritizes covering uncovered towers, and the total coverage does not exceed the maximum coverage limit of a single cell. Covered towers can be used as redundant coverage to ensure that the coverage calculation meets the actual constraints.

[0065] Step 2: Objective function calculation: Calculate the objective function value of the new solution based on the effective coverage matrix;

[0066] Step 3: Simulated annealing acceptance criteria: Calculate the difference Δ between the objective function of the new solution and the current solution. If Δ > 0, directly accept the new solution and update the global optimal solution; if Δ ≤ 0, accept it with probability exp(Δ / T) (T is the current temperature) to avoid the algorithm getting trapped in local optima.

[0067] Finally, the operator weights are dynamically adjusted and the program is iteratively terminated. The specific process is as follows:

[0068] Step 1: Temperature Update: A segmented cooling strategy is adopted, with the first 70% of iterations cooling at a faster rate and the last 30% of iterations cooling at a slower rate, balancing the algorithm's exploration and convergence speed.

[0069] Step 2: Operator weight update: The operator weight is updated every 10 iterations, and adjusted based on the operator's average score (total score / number of times used) to ensure that well-performing operators are used more often, while avoiding negative weights;

[0070] Step 3: Iteration Termination: When the number of iterations reaches the preset maximum number of iterations, the algorithm terminates and outputs the global optimal solution.

[0071] Compared with the prior art, the beneficial effects of the present invention are:

[0072] This invention addresses the shortcomings of existing optimization studies on fixed UAV nest location models for power line inspection needs by proposing an optimization method based on GIS-ALNS for such models. Compared to traditional precise and heuristic algorithms, this invention uses ArcGIS to build a suitability raster map, considers tower coverage and nest service performance, constructs a multi-objective optimization location model, and then employs the Adaptive Large Neighborhood Search (ALNS) algorithm to solve the multi-objective optimization location model. This method effectively avoids the limitations of local optima, achieves a dynamic balance between coverage effect and location quality, makes the location scheme for power line inspection nests more scientific and reasonable, and thus significantly improves the overall efficiency of power line inspection and reduces the comprehensive cost throughout the project's lifecycle. Attached Figure Description

[0073] Figure 1 This is a flowchart of the method of the present invention.

[0074] Figure 2 This is a map showing the distribution of some power poles and no-fly zones in a certain area.

[0075] Figure 3 This is a grid map showing the suitability of nesting facility construction in a certain region.

[0076] Figure 4 This is a raster map of candidate points selected in the initial screening.

[0077] Figure 5 This is a diagram showing the nest location results obtained using this method.

[0078] Figure 6 This is a graph showing the change in the objective function and coverage of the random nest number when using this method.

[0079] Figure 7 This is a diagram showing the nest location results obtained using a greedy algorithm.

[0080] Figure 8 This is a graph showing the change in the number of random nests using the objective function of a greedy algorithm. Detailed Implementation

[0081] The present invention will be further described in detail below through embodiments, but it should not be construed as limiting the scope of the subject matter of the present invention to the following embodiments. All technologies implemented based on the above content of the present invention fall within the scope of the present invention.

[0082] This invention proposes an optimization method for the fixed UAV nest location model based on GIS-ALNS for power line inspection needs, as shown in the flowchart below. Figure 1 As shown, it includes the following steps:

[0083] (1) Considering the differences in suitability of nest construction in different regions, a suitability raster map was created using ArcGIS to initially screen sites with higher suitability.

[0084] The specific steps in step (1) include:

[0085] 1.1 Obtain raster data such as slope, aspect, land use type, and population density of the area surrounding the pole tower, as well as road vector data.

[0086] 1.2 The suitability of different regions is defined by five different attributes: topographic features, transportation convenience, population density, land use type, and whether they are located in a no-fly zone.

[0087] The classification criteria for slope suitability are based on angle ranges, as detailed below:

[0088] 1 point (suitable height): The slope is between 0 and 5 degrees;

[0089] 3 / 4 (suitable): Slope between 5 and 10 degrees;

[0090] 2 / 4 (generally suitable): Slope between 10 and 15 degrees;

[0091] 1 / 4 (lower suitable): The slope is in the range of 15-20 degrees;

[0092] 0 points (Unsuitable): Slope is above 20 degrees.

[0093] Slope aspect suitability classification is based on azimuth range:

[0094] 1 point (suitable height): Slope facing south, corresponding to an azimuth of 90-270°;

[0095] 0 points (unsuitable): Slope facing north, corresponding to an azimuth of 0-90° or 270-360°.

[0096] Traffic suitability classification is based on the Euclidean distance range (in meters) from the road:

[0097] 1 point (suitable height): 100-500 meters from the road;

[0098] 3 / 4 (Suitable): 500-1100 meters from the road;

[0099] 2 / 4 (Generally suitable): 1100-1500 meters from the road;

[0100] 1 / 4 (lower suitable): 0-100 meters or 1500-2100 meters from the road;

[0101] 0 points (Unsuitable): More than 2100 meters away from the road.

[0102] Population density suitability classification is based on the population size per unit area (unit: people):

[0103] 1 point (Highly suitable): Population size 0-50 people;

[0104] 3 / 4 (more suitable): Population size 50-300 people;

[0105] 2 / 4 (Generally suitable): Population size 300-1000 people;

[0106] 1 / 4 (lower suitable): Population size 1000-2000 people;

[0107] 0 points (Unsuitable): Population exceeds 2,000.

[0108] The suitability classification criteria for no-fly zones are based on whether the area is located within a no-fly zone:

[0109] 1 point (suitable altitude): The area is located outside the no-fly zone;

[0110] 0 points (Unsuitable): The area is located within a no-fly zone.

[0111] The land use suitability classification standard is based on land use type:

[0112] 1 point (suitable height): Bare ground;

[0113] 3 / 4 (most suitable): Artificial surface;

[0114] 2 / 4 (generally suitable): Grassland;

[0115] 1 / 4 (lower suitability): Woodland;

[0116] 0 points (Unsuitable): Farmland, water bodies or wetlands.

[0117] 1.3. The Analytic Hierarchy Process (AHP) and the entropy weighting method are used to assign weights to each attribute. The suitability of a single grid is obtained by weighted summation of the suitability of all attributes.

[0118] 1.4 Select the top 50% of grids in terms of construction suitability as preliminary candidate sites for nest construction.

[0119] (2) Taking into account both the coverage requirements of the poles and the service performance of the cell towers, design an objective function with a penalty for exceeding the coverage limit of a single cell tower.

[0120] In step (2), an objective function with a single-nest coverage over-limit penalty is calculated. To ensure that the number of towers covered by each nest is within its performance support range, a single-nest coverage over-limit penalty is defined: for each selected nest location, if the number of nests it covers exceeds the limit, a penalty of -1 is used; the objective function with the single-nest coverage over-limit penalty... as follows:

[0121] in, B represents the coverage rate weight, where B is the tower coverage rate. Let A be the average nest suitability weight, γ be the average nest suitability, γ be the penalty coefficient, and P be the cumulative penalty for selecting nests that exceed the limit. The formula for calculating P is as follows:

[0122]

[0123] Where, x j Indicates whether to select candidate nest c j , This is an indicator function; if the nest is selected... If the number of covered poles exceeds the limit, return 1; otherwise, return 0.

[0124] (3) Considering the limitations of UAV inspection radius and maximum number of UAV nests under real conditions, a multi-objective optimization site selection model is constructed with the goal of maximizing the coverage of power poles and the highest average suitability of UAV nests.

[0125] In step 3, considering limitations such as the drone inspection radius and the maximum number of drone nests under realistic conditions, the goal is to maximize the coverage of power poles and the average suitability of drone nests, as shown in the following formula:

[0126] Max

[0127] in

[0128]

[0129] z i Indicates tower e i Whether it is covered by at least one nest, where E represents the set of poles, C represents the set of candidate nests, and s j Indicates nest c j suitability, y ij Indicates tower e i Is it nested? j Coverage, p j Indicates nest c j The penalty received for exceeding the limit.

[0130] The constraints are specifically manifested as follows:

[0131]

[0132] Where, d ij Indicates tower e i With Nest C j The Euclidean distance between them This is an indicator function, which is 1 when the condition is true and 0 otherwise. K is the maximum number of nests.

[0133] (4) The adaptive large neighborhood search (ALNS) algorithm is used to solve the multi-objective optimization location model.

[0134] The specific steps in step (4) include:

[0135] 4.1 Generating the initial solution:

[0136] Step 1: Generate initial solution using a greedy strategy: initialize the nest selection state to "not selected" and the tower coverage state to "not covered", and pre-calculate the initial coverage capability of each candidate nest;

[0137] Step 2: Iterative selection of candidate cellars: Each time, from the candidate cellars that are "not selected but still have coverage capacity", select the location with the most uncovered towers, mark it as "selected" and update the global coverage status. At the same time, dynamically reduce the remaining coverage capacity of other candidate cellars until there are no uncovered towers or the maximum number of cellars is reached.

[0138] 4.2 Execute the ALNS core iteration process:

[0139] Step 1: Operator selection: Based on the roulette wheel method, randomly select the destroying operator and the repair operator according to the current operator weight (the higher the weight of the operator, the greater the probability of being selected), and update the operator usage count;

[0140] Step 2: Destruction Phase: Invoke the selected destruction operator to "destroy" the current solution and generate a partial solution;

[0141] Step 3: During the repair phase, the selected repair operator is invoked to "repair" the partial solution and replenish the nests to the maximum number limit.

[0142] 4.3, Solution Evaluation and Acceptance:

[0143] Step 1: Effective Coverage Calculation: For the repaired new solution, calculate the effective coverage matrix - traverse the selected cell in descending order of suitability, each cell prioritizes covering uncovered towers, and the total coverage does not exceed the maximum coverage limit of a single cell. Covered towers can be used as redundant coverage to ensure that the coverage calculation meets the actual constraints.

[0144] Step 2: Objective function calculation: Calculate the objective function value of the new solution based on the effective coverage matrix;

[0145] Step 3: Simulated annealing acceptance criteria: Calculate the difference Δ between the objective function of the new solution and the current solution. If Δ > 0, directly accept the new solution and update the global optimal solution; if Δ ≤ 0, accept it with probability exp(Δ / T) (T is the current temperature) to avoid the algorithm getting trapped in local optima.

[0146] 4.4 Dynamic Adjustment of Operator Weights and Termination of Iteration:

[0147] Step 1: Temperature Update: A segmented cooling strategy is adopted, with the first 70% of iterations cooling at a faster rate and the last 30% of iterations cooling at a slower rate, balancing the algorithm's exploration and convergence speed.

[0148] Step 2: Operator weight update: The operator weight is updated every 10 iterations, and adjusted based on the operator's average score (total score / number of times used) to ensure that well-performing operators are used more often, while avoiding negative weights;

[0149] Step 3: Iteration Termination: When the number of iterations reaches the preset maximum number of iterations, the algorithm terminates and outputs the global optimal solution.

[0150] The technical solution of the present invention will be further illustrated below with a specific embodiment.

[0151] Power poles in a portion of Jiangsu Province were selected as the research subject. This region has 1499 power poles, of which 177 are located within a no-fly zone. A drone took off from its nest, with a patrol radius of 3000 meters, and inspected the power poles within its coverage area. The distribution of the power poles is as follows: Figure 2 As shown.

[0152] (a) Candidate point screening

[0153] Raster data and road vector data for the region, including slope, aspect, population density, and land use type, were acquired. The weights of each attribute were obtained using the analytic hierarchy process (AHP) and entropy weighting. Based on suitability classification criteria, the pixel values ​​of the raster data were reclassified to their corresponding suitability levels. According to these criteria, multiple buffer zones were established around the road vector data. The buffer vector data was converted to raster data, and its pixel values ​​were reclassified to their corresponding suitability levels. Finally, based on the weights of each attribute, a weighted sum of all attributes was calculated to obtain a suitability raster map of the region, as shown below. Figure 3 As shown in the image, the grid cells with the top 50% suitability are extracted to obtain a preliminary candidate point map, as shown in the image. Figure 4 As shown

[0154] (2) Solving with ALNS algorithm

[0155] To maximize power pole coverage and average drone nest suitability, a multi-objective optimization model for site selection is constructed, with limitations set on drone inspection radius and maximum number of drone nests. This model imports pole coordinate data and corresponding suitability weights, and uses the ALNS algorithm and a greedy algorithm to solve the objective function. The effective coverage and suitability of each drone nest under different numbers are obtained, as shown below. Figure 5 , 6 and Figure 7 , 8 As shown, by introducing a suitability factor and further optimizing the initial solution in the ALNS algorithm, the tower coverage rate is 2%-5% higher than that of the greedy algorithm under a limited number of cell towers. The results indicate that the model considering suitability requirements proposed in this invention has a better ability to meet the target requirements and can effectively optimize the fixed cell tower location decision.

[0156] (3) Results of fixed machine nest site selection

[0157] The site selection was carried out, and the results were as follows: Figure 5 As shown.

[0158] The optimal site selection schemes shown demonstrate the algorithm's ability to generate efficient and tailored solutions based on the specific characteristics and constraints of each scenario. The results of this case study highlight the significant potential of adaptive large-area search algorithms in coordinating and optimizing fixed tower location schemes. By implementing adaptive decision-making, this method can effectively maximize tower coverage, reduce construction and maintenance costs, and improve overall inspection performance in practical applications.

[0159] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent substitutions, and improvements made by those skilled in the art to the above embodiments without departing from the scope of the technical solution of the present invention, based on the technical essence of the present invention, shall still fall within the protection scope of the technical solution of the present invention.

Claims

1. A GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs, characterized by, The method comprises the following steps: Step 1: considering the difference of nest construction suitability in different regions, using ArcGIS to establish a suitability grid map, and preliminarily screening sites with high suitability; Step 2: comprehensively considering the tower coverage demand and the service performance of the nest, a target function with single-nest coverage overrun penalty is designed; Step 3: considering the limitation of the inspection radius of the unmanned aerial vehicle and the maximum number of nests under the actual condition, a multi-objective optimization site selection optimization model is constructed with the maximum power tower coverage rate and the highest average suitability of nests as the target; Step 4: based on the target function, the adaptive large neighborhood search (ALNS) algorithm is used to solve the multi-objective optimization site selection model.

2. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 1, characterized in that, The specific process of step 1 is as follows: Step 1.1: obtain the grid data of slope, slope direction, land use type and population density, and road vector data in the surrounding area of the tower; Step 1.2: considering the attribute differences of different regions in terms of terrain characteristics, traffic convenience, population density, land use type and whether in the no-fly zone, define the suitability of different attributes; Step 1.3: use the analytic hierarchy process and entropy weight method to weight each attribute, and then obtain the construction suitability of the grid by weighting and summing the suitability of all attributes of the single grid; Step 1.4: select the top 50% of the construction suitability grid as the preliminary screened nest construction candidate point.

3. The GIS-ALNS-based optimization method for a fixed nest site selection model for power inspection demand-oriented UAVs according to claim 2, characterized in that, In step 1.2, the suitability of five attributes is defined as follows: The slope suitability classification standard is based on the angle interval, which is as follows: 1 point for high suitability: the slope is in the range of 0-5 degrees; 3 / 4 points for relatively suitable: the slope is in the range of 5-10 degrees; 2 / 4 points for general suitability: the slope is in the range of 10-15 degrees; 1 / 4 points for relatively low suitability: the slope is in the range of 15-20 degrees; 0 points for unsuitable: the slope is more than 20 degrees; The slope direction suitability classification is based on the range of azimuth angle: 1 point for high suitability: the slope direction is south, corresponding to the azimuth angle of 90-270°; 0 points for unsuitable: the slope direction is north, corresponding to the azimuth angle of 0-90° or 270-360°; The traffic suitability classification standard is based on the Euclidean distance interval from the road: 1 point for high suitability: the distance from the road is 100-500 meters; 3 / 4 points for relatively suitable: the distance from the road is 500-1100 meters; 2 / 4 points for general suitability: the distance from the road is 1100-1500 meters; 1 / 4 points for relatively low suitability: the distance from the road is 0-100 meters or 1500-2100 meters; 0 points for unsuitable: the distance from the road is more than 2100 meters; The population density suitability classification is based on the population number per unit area: 1 point for high suitability: the population number is 0-50; 3 / 4 points for relatively suitable: the population number is 50-300; 2 / 4 points for general suitability: the population number is 300-1000; 1 / 4 points for relatively low suitability: the population number is 1000-2000; 0 points for unsuitable: the population number is more than 2000; The no-fly zone suitability classification standard is based on whether the region is in the no-fly zone: 1 point for high suitability: the region is outside the no-fly zone; 0 points for unsuitable: the region is in the no-fly zone; The land use suitability classification standard is based on the land use type: 1 point for highly suitable: bare land; 3 / 4 points for less suitable: artificial surface; 2 / 4 points for general suitable: grassland; 1 / 4 points for less suitable: forest land; 0 points for unsuitable: farmland, water body or wetland.

4. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 1, characterized in that, The objective function with single machine nest coverage over-limit penalty in step 2 is as follows: To ensure that the number of towers covered by each machine nest is within its performance support range, define a single machine nest coverage over-limit penalty: for each selected machine nest site, if the number of machine nests it covers exceeds the limit, use a penalty of -1; Objective function with single machine nest covering over-limit penalty As follows: ； wherein, B is the tower coverage rate, A is the average suitability of the nest, γ is the penalty coefficient, P is the cumulative penalty of the selected nest exceeding the limit, and the calculation formula of P is as follows: ； where x j represents whether the candidate nest c j is selected is an indicator function that returns 1 if the number of towers covered by the candidate nest c j exceeds the limit, and 0 otherwise.

5. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 1, characterized in that, In step 3, the multi-objective optimization site selection optimization model is constructed as follows: Considering the limitations of the inspection radius of the unmanned aerial vehicle and the maximum number of machine nests under real conditions, the maximum power tower coverage rate and the highest average suitability of machine nests are taken as the objectives, and the formula is as follows: Max ; Wherein ； ； ； ； ； z i represents a tower e i is covered by at least one nest, E represents a set of towers, C represents a set of candidate nests, s j represents a nest c j fitness, y ij represents a tower e i is covered by a nest c j p j represents a nest c j received over-limit penalty; The constraint conditions are as follows: ； ； ； where d ij represents the Euclidean distance between the tower e i and the nest c j , is an indicator function, which is 1 if the condition holds, and 0 otherwise, and K is the maximum number of nests.

6. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 1, characterized in that, The specific process of step 4 is as follows: Step 4.1: generate an initial solution; Step 4.2: execute the ALNS core iteration process; Step 4.3: solution evaluation and acceptance; Step 4.4: dynamic adjustment of operator weight and iteration termination.

7. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 6, characterized in that, The specific process of step 4.1 is as follows: Step 1: generate an initial solution using a greedy strategy: initialize the machine nest selection state to "not selected" and the tower coverage state to "not covered", and pre-calculate the initial coverage capacity of each candidate machine nest; Step 2: iteratively select candidate machine nests: each time, select the site with the most uncovered towers from the "not selected and still has coverage capacity" candidate machine nests, mark it as "selected" and update the global coverage state, while dynamically decreasing the remaining coverage capacity of other candidate machine nests, until there are no uncovered towers or the maximum number of machine nests is reached.

8. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 6, characterized in that, The specific process of step 4.2 is as follows: Step 1: operator selection: based on the roulette method, randomly select the destruction operator and the repair operator according to the current operator weight, the higher the weight, the greater the probability of being selected, and update the operator usage count; Step 2: destruction phase: call the selected destruction operator to "destroy" the current solution and generate a partial solution; Step 3: call the selected repair operator to "repair" the partial solution and supplement the machine nests to the maximum number limit.

9. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 6, characterized in that, The specific process of step 4.3 is as follows: Step 1: effective coverage calculation: for the repaired new solution, calculate the effective coverage matrix, and traverse the selected machine nests in descending order of suitability, each machine nest preferentially covers the uncovered towers, and the total number of coverage does not exceed the single machine nest maximum coverage limit, the covered towers can be used as redundant coverage, to ensure that the coverage calculation conforms to the actual constraints; Step 2: target function calculation: based on the effective coverage matrix, calculate the target function value of the new solution; Step 3: simulated annealing acceptance criterion: calculate the difference Δ between the target function values of the new solution and the current solution, if Δ>0, directly accept the new solution and update the global optimal solution; if Δ≤0, then accept it with a probability of exp(Δ / T), where T is the current temperature, to avoid the algorithm falling into local optimum.

10. The GIS-ALNS-based optimization method for a UAV fixed nest site selection model for power inspection needs according to claim 6, characterized in that, The specific process of step 4.4 is as follows: Step1: Temperature update: Adopt a segmented cooling strategy, the first 70% of the iteration times are cooled at a faster rate, and the last 30% of the iteration times are cooled at a slower rate, balancing the exploration and convergence speed of the algorithm; Step2: Operator weight update: Update the operator weight every 10 iterations, based on the average score adjustment of the operator, to ensure that good performing operators are used more, while avoiding negative weights; Step3: Iteration termination: When the number of iterations reaches the pre-set maximum number of iterations, terminate the algorithm and output the global optimal solution.

Images