A power transmission and transformation project intelligent site selection and line selection method, system, device and medium

Abstract

The application discloses a power transmission and transformation engineering intelligent site selection and line selection method, system, device and medium, relates to the technical field of power system planning, and comprises the following steps: constructing a grid model by fusing multi-source data, generating a passing cost map, and screening out candidate substation positions; encoding the stations and paths into schemes; constructing guide information for path nodes and calculating fitness in iteration; using an elite solution set to guide population evolution; combining a local disturbance strategy to perform iteration until a convergence condition is met; and outputting a final Pareto frontier solution set. Through the synergistic effect of multi-source data fusion and cost modeling, station-line unified coding and wind field guided search, multi-objective elite optimization and local disturbance repair mechanism, the application realizes efficient generation of a power transmission and transformation station-line integrated scheme under complex constraints, and significantly improves the scientificity, efficiency and engineering practicability of planning.

Description

Technical Field

[0001] This invention relates to the field of power system planning technology, specifically to a method, system, equipment, and medium for intelligent site selection and route selection in power transmission and transformation projects. Background Technology

[0002] Site selection and route selection for power transmission and transformation projects are crucial aspects of power system planning. The quality of these decisions directly impacts the economic efficiency, environmental friendliness, and long-term operational reliability of the power grid. With the rapid development of geographic information systems, remote sensing, and intelligent optimization algorithms, traditional design models relying on human experience and qualitative analysis are gradually evolving towards data-driven, quantitative optimization, and intelligent approaches. Current mainstream technologies typically employ staged processing or integrate classic swarm intelligence algorithms for path optimization. By setting penalty functions for spatial constraints such as terrain and ecology, and constructing single-objective or simple weighted multi-objective functions primarily based on path length or economic cost for optimization, planning efficiency and feasibility of solutions are improved to some extent, providing a preliminary technical framework for automated route selection.

[0003] However, existing technologies still face significant limitations when dealing with complex real-world engineering scenarios involving high resolution and multiple constraints: First, in terms of spatial constraint modeling, most methods rely on static penalty functions to handle infeasible regions, lacking effective utilization of continuous geographic gradient information. This often leads to unreasonable abrupt changes in the generated paths or conflicts with high-cost regions, resulting in poor physical feasibility. Second, in problem modeling, substation site selection and line route planning are generally treated as two independent, sequential stages, ignoring the strong coupling between them. This can easily lead to the overall solution falling into local optima and cannot guarantee the overall optimality of station-line coordination. Finally, at the optimization algorithm level, traditional swarm intelligence algorithms often suffer from slow convergence speed, uneven distribution of Pareto fronts, and premature convergence when dealing with high-dimensional multi-objective problems. Furthermore, few mechanisms can dynamically integrate geographic information and historical elite solutions to guide the search direction, resulting in insufficient search efficiency and stability of the algorithm in complex solution spaces. Summary of the Invention

[0004] In view of the above-mentioned existing problems, the present invention provides a method, system, equipment and medium for intelligent site selection and line selection in power transmission and transformation projects, in order to solve the problems of lack of effective utilization of continuous geographical gradient information, the overall scheme getting stuck in local optima, the inability to guarantee the overall optimality of station-line coordination, and the insufficient search efficiency and stability of the algorithm in complex solution space in the existing technology.

[0005] To address the aforementioned technical problems, a smart site selection and route selection method for power transmission and transformation projects is proposed, including: Multi-source spatial data is acquired and fused to construct a unified raster spatial model. Based on the raster spatial model, the passage cost value for each raster cell is calculated, generating a passage cost map. Based on the spatial data, candidate substation locations that meet preset construction conditions are identified and screened. The candidate locations are comprehensively scored according to evaluation indicators, and the candidate location set is determined based on the scoring results. The selected substation locations and connection paths are uniformly coded to form a station-line scheme. In the optimization iteration, the path nodes of each individual are updated to construct guiding information, and the fitness value of the optimization objective is calculated for each individual. The objective values ​​are normalized. With the goal of optimizing the fitness of multiple objectives, the population is iteratively updated. An elite solution set storing non-dominated solutions is introduced to guide the population evolution direction, and a local perturbation strategy is combined until the convergence condition is met. When the termination condition is met, the Pareto front solution set is output.

[0006] As a preferred embodiment of the intelligent site selection and route selection method for power transmission and transformation projects described in this invention, the generation of the travel cost map includes: integrating geographic information data within the study area and uniformly converting it into the same raster format; establishing an integrated spatial representation model; calculating the travel cost value for each grid cell in the rasterized space; and determining the final value by setting different weight coefficients for the cost, thereby generating the cost map.

[0007] As a preferred embodiment of the intelligent site selection and route selection method for power transmission and transformation projects described in this invention, the step of determining the candidate location set based on the scoring results includes: automatically analyzing and processing image information in multi-source spatial data, extracting a probability distribution map characterizing the suitability of geographical areas for construction, and preliminarily delineating highly suitable areas based on the probability distribution map by setting a threshold. For each initially identified region, a multi-indicator quantitative evaluation is conducted, and the comprehensive score of each region is calculated based on a weighted scoring function; Based on the overall score, a predetermined number of regional center points with the highest ranking are selected to form a set of candidate substation locations for entering the optimization process.

[0008] As a preferred embodiment of the intelligent site selection and route selection method for power transmission and transformation engineering described in this invention, the construction of guidance information includes: querying the travel cost map for nodes to be updated in the path sequence, calculating the rate of change of cost between the grid where the current node is located and the adjacent grids, and determining local gradient information indicating the direction of cost reduction. Calculate the unit direction vector from the current node to the end substation location of its path, as the direction information, and synthesize the local gradient information and the direction information into a vector according to the preset weight coefficients to generate a composite vector, which serves as the guiding information for the current node to move in this iteration.

[0009] As a preferred embodiment of the intelligent site selection and route selection method for power transmission and transformation projects described in this invention, the fitness value of the calculated optimization target includes the calculated path length target, ecological impact target, construction cost target, and power supply reliability target, and the original values ​​of the four targets are normalized and converted into dimensionless fitness scores. The objectives for calculating path length include: calculating the total extension length of the path encoded by an individual across all grid cells; calculating ecological impact includes: identifying the ecological sensitivity score of each grid cell traversed by the path, multiplying the length of the path segments within each grid cell by the ecological sensitivity score of the current grid cell, and summing the results to obtain the ecological impact-weighted total length; calculating construction cost includes: calculating the slope of the area traversed by the path based on digital elevation model data, and combining this with the path length to calculate an estimate of the total construction cost using a comprehensive cost function; and calculating power supply reliability includes: assessing the topological connection distance between the substation location at the path's endpoint and key nodes in the existing main power grid to quantify the level of power supply reliability.

[0010] As a preferred embodiment of the intelligent site selection and line selection method for power transmission and transformation engineering described in this invention, the guiding population evolution direction includes: after each generation iteration, performing non-dominated sorting on all current individuals, and storing all non-dominated solutions of the first level of sorting into a fixed-size storage structure in the elite memory pool. The update rules for the elite memory pool include: when a newly generated solution dominates an old solution in the pool, the new solution replaces the old solution; when the new solution does not dominate any of the solutions in the pool, the crowding degree between the new solution and the solutions in the pool is calculated; when the target space region where the new solution is located is sparser than the solutions in the pool, the new solution replaces the solution with the highest current crowding degree. The formula for calculating the crowding degree of the new solution and the solutions in the pool is expressed as: in, Let be the crowding level of the i-th individual. Let i+1 be the normalized fitness value of the i+1th individual on the kth optimization objective dimension. Let be the normalized fitness value of the (i-1)th individual on the k-th optimization objective dimension. This represents the maximum fitness value in the k-th objective dimension among all current individuals. It represents the minimum fitness value in the k-th target dimension among all current individuals. To optimize the total number of objectives, To optimize the target index, For individual indexes; The formula for individual renewal in an elite population is expressed as: in, Let i be the current position of the i-th individual in the t-th iteration. Let i be the new position of the i-th individual after the (t+1)-th iteration. The wind field propulsion coefficient, This is the wind field guiding vector. This represents the local disturbance intensity coefficient. It is a random perturbation vector; The local perturbation strategy includes, after each iteration of population position update, traversing the path node sequence of each individual with a preset probability, superimposing a Gaussian distributed random offset with zero mean and controllable variance on the grid coordinates of the selected nodes, and after the perturbation update, checking whether all path nodes of each individual fall into infeasible grids. For any node that falls into an infeasible grid, adjusting the coordinates to the center of the nearest feasible grid within the surrounding eight neighborhoods and with a passage cost lower than a preset feasibility threshold. The local perturbation operator formula is expressed as: in, Let l be the position of the l-th node in the path during the t-th iteration. This represents the new position of the current path node after perturbation in the (t+1)th iteration. As a disturbance factor, To have a mean of zero and a variance of A random variable with a Gaussian distribution.

[0011] As a preferred embodiment of the intelligent site selection and line selection method for power transmission and transformation engineering described in this invention, the iterative update of the population includes: for each individual in the population, taking the composite vector as the main guiding direction, the introduced elite guiding vector as the auxiliary direction, and the introduced random offset as the exploration term, calculating the incremental change of the position through a weighted summation formula, and superimposing the increment onto the current position to generate a new generation of individual positions; The elite guiding vector is represented as: in, For individual location The wind field steering vector when the path node is located at grid coordinates (a, b). This represents the negative direction of the grid cost gradient vector at the current position. The target guiding coefficient, This represents the direction vector of the elite individual selected from the elite memory pool; The formula for adjusting the update direction under the guidance of elites is expressed as follows: in, This is the adjusted composite direction vector used to update ordinary individuals. For balance coefficient, This represents the current position of the j-th individual that is currently being updated; The convergence conditions include reaching the preset maximum number of iterations, outputting the final solution set, and extracting all individuals within the first non-dominated level after performing non-dominated sorting.

[0012] The beneficial effects of this preferred technical solution are as follows: by introducing an elite memory pool based on congestion management, the population is effectively guided to evolve towards the Pareto front. Combined with local perturbation and feasibility repair strategies, the diversity of solutions and engineering feasibility are ensured, while balancing the development and exploration capabilities of the algorithm. This significantly improves the optimization efficiency and solution quality, and outputs a uniformly distributed and directly applicable Pareto optimal station line scheme, providing strong scientific support for multi-objective decision-making in power transmission and transformation projects under complex geographical conditions.

[0013] As a preferred embodiment of the intelligent site selection and route selection system for power transmission and transformation projects described in this invention, it is characterized by including a spatial data fusion and cost modeling module, a candidate site intelligent screening module, a problem unified coding and modeling module, a geographic gradient guidance and path search module, and a multi-objective collaborative optimization and decision output module.

[0014] The spatial data fusion and cost modeling module is used to integrate multi-source heterogeneous geospatial data from remote sensing, topography, land use and ecological protection areas. It constructs a digital spatial base through unified coordinates and rasterization processing, and uses a preset cost function to calculate the cost value of each raster unit, which comprehensively reflects ecological sensitivity and engineering economic costs, to generate a cost map.

[0015] The intelligent candidate site screening module is used to automatically analyze remote sensing images using deep learning image segmentation technology, identify and extract areas with construction conditions, generate a site suitability probability map, and combine multi-dimensional engineering evaluation indicators such as load center distance, geological stability, and traffic accessibility to comprehensively quantify and rank candidate areas, and output a set of candidate substation locations.

[0016] The unified coding and modeling module is used to encode the spatial coordinates of the substation and the sequence of transmission line path nodes connecting the current station in an integrated manner to form a complete solution expression, and to adjust the station location and path direction simultaneously.

[0017] The geographic gradient guidance and path search module is used to dynamically calculate the guidance vector for each node to be adjusted on the path during the optimization iteration process, transforming geographic environmental information into a search driving force with physical meaning.

[0018] The multi-objective collaborative optimization and decision output module is used to construct a multi-objective fitness function evaluation scheme that includes path length, ecological impact, cost and reliability, adopt an elite retention strategy to maintain high-quality solutions, combine non-dominated sorting and crowding calculation to maintain the diversity and distribution of solutions, introduce local perturbation and feasibility repair mechanisms to handle spatial hard constraints, and output a Pareto optimal solution set that achieves the best trade-off on multiple objectives.

[0019] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of a method for intelligent site selection and line selection in power transmission and transformation engineering.

[0020] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for intelligent site selection and line selection in power transmission and transformation engineering.

[0021] The beneficial effects of this invention are as follows: By fusing multi-source spatial data and constructing a rasterized access cost map, this invention provides a high-precision, quantifiable geographical constraint model for the optimization process, laying the data foundation for refined optimization; by adopting a unified coding mechanism for station sites and path nodes, it breaks the limitations of phased optimization and realizes collaborative global optimization of substation site selection and transmission line paths; by integrating local cost gradients and target direction information through a wind field guidance mechanism, it provides adaptive and directional intelligent drive for path search, improving search efficiency and the physical rationality of paths; by constructing a multi-objective fitness function and combining an elite memory pool, non-dominated sorting, and congestion maintenance strategy for population evolution, it ensures that the algorithm can efficiently and stably converge to a uniformly distributed Pareto optimal solution set in complex multi-objective spaces; and by introducing local perturbation and automatic feasibility repair mechanisms, it enhances the algorithm's ability to escape local optima and ensures that all solutions satisfy hard spatial constraints, significantly improving the decision-making quality, efficiency, and scientific nature of power transmission and transformation engineering planning. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating an intelligent site selection and route selection method for power transmission and transformation projects, provided as an embodiment of the present invention.

[0024] Figure 2The present invention provides a system scheme flowchart for an intelligent site selection and route selection system for power transmission and transformation projects according to an embodiment of the present invention. Detailed Implementation

[0025] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0026] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for intelligent site selection and route selection in power transmission and transformation projects is provided, comprising: S100: Acquire multi-source spatial data and perform fusion processing to construct a unified grid spatial model. Based on the grid spatial model, calculate the passage cost value for each grid cell, generate a passage cost map, and identify and screen candidate substation locations that meet the preset construction conditions based on the spatial data. Perform comprehensive scoring on the candidate locations according to the evaluation indicators, and determine the candidate location set based on the scoring results.

[0027] S200: The selected substation locations and connection paths are uniformly coded to form a station-line scheme. In the optimization iteration, the path nodes of each individual are updated to construct guidance information, and the fitness value of the optimization target is calculated for each individual. The target values ​​are then normalized.

[0028] S300: With the goal of optimizing multi-objective fitness, the population is iteratively updated. An elite solution set that stores non-dominated solutions is introduced to guide the direction of population evolution. A local perturbation strategy is combined until the convergence condition is met. When the termination condition is met, the Pareto front solution set is output.

[0029] It should be noted that this invention achieves efficient and scientific optimization under complex geographical constraints by constructing a refined grid cost model and a multi-dimensional target evaluation system, and combining it with a physically meaningful guidance mechanism. Furthermore, by introducing an elite memory pool and a balanced evolutionary strategy, the global convergence and solution set diversity of the algorithm are effectively guaranteed, and the final output is a high-quality, evenly distributed, and engineering-feasible Pareto optimal solution.

[0030] Example 2, refer to Figure 1 This is a second embodiment of the present invention, which provides a method for intelligent site selection and route selection in power transmission and transformation projects, including: In this embodiment of the application, step S100, the construction of a unified grid space model includes steps S101 to S103: S101: Integrate multi-source spatial data of high-resolution remote sensing images, digital elevation models, land use types, and ecological protection red lines within the study area, convert them all to raster format using a unified coordinate system, and establish a comprehensive spatial representation model.

[0031] S102: Calculate the passage cost for each grid cell in the model. This value comprehensively quantifies the terrain and ecological costs that the route must bear when passing through that location. Specifically, this cost is jointly determined by a normalized ecological sensitivity score and a construction cost factor composed of factors such as slope, geology, and land use, and the trade-off between the two is adjusted by preset weights. The formula for calculating the cost of passage is expressed as: in, Let (a, b) be the passage cost of the grid cell. The ecological sensitivity score for grid cell (a, b) is given. Infinite The weighting coefficients for ecological sensitivity scoring. The construction cost factor for grid cell (a, b) Weighting coefficients for construction cost factors S103: The system traverses all grid cells to generate a global travel cost map, which is used for comprehensive evaluation of subsequent paths and serves as a key gradient information source for the wind field guidance mechanism in the optimization algorithm.

[0032] In an optional implementation, in step S100, the construction of a unified grid spatial model further includes identifying and labeling various types of high-cost source features, defining a cost attenuation model (exponential attenuation) for each type of source feature, that is, the farther away from the source feature, the lower the impact cost, and for any grid in the space, calculating the superposition value of the impact of all cost sources as the final passage cost of the current grid.

[0033] In another optional implementation, in step S100, the construction of a unified grid spatial model may further include collecting historical route samples, rasterizing each route, extracting multidimensional geographical features (elevation, vegetation index, geological code, etc.) of each grid as input, using the actual unit length comprehensive cost (after conversion) of the route at the grid as training labels, and training the model using random forest or neural network algorithms. For new area planning, the features of the corresponding grid are input, and the model can predict the passage cost.

[0034] Furthermore, in this embodiment of the application, in step S100, identifying and screening candidate substation locations that meet the preset construction conditions includes steps S111~S113: S111: Deep feature extraction is performed on high-resolution remote sensing images using a semantic segmentation model based on U-Net. A construction probability map is generated by processing the images through convolution operations and the Sigmoid activation function. A threshold of 0.7 is used to perform binarization to extract highly feasible construction area patches. The formula for the semantic segmentation model is expressed as: in, For input high-resolution remote sensing image patches, The output is a site construction probability map, where each pixel value represents the probability that the current location is suitable for site construction. The convolution kernel weight matrix is... This is the convolution kernel bias vector. This is the Sigmoid activation function.

[0035] S112: Construct a quantitative evaluation system for each candidate area that includes multiple indicators such as distance to the load center, geological stability, and traffic accessibility, and calculate the comprehensive score of each site by weighted summation; The formula for calculating the comprehensive score of each site is expressed as follows: in, The comprehensive score for the location of the g-th candidate substation is... This is an index for the location of candidate substations. The total number of evaluation indicators. For evaluation index, Let m be the weight coefficient of the m-th evaluation indicator. For the m-th evaluation index with respect to the coordinates of the g-th station site The scoring function, Let be the planar coordinates of the location of the g-th candidate substation.

[0036] S113: Sort all candidate areas according to the scoring results, select the center point or optimal point of the top A area to form a high-quality initial set of candidate substation locations, providing accurate input for the optimization algorithm.

[0037] In an optional implementation, in step S100, identifying and screening candidate substation locations that meet preset construction conditions further includes performing multi-scale segmentation on the remote sensing image, dividing the image into homogeneous regions, extracting multiple features of each region, such as spectral features (NDVI vegetation index, water index), texture features, and shape features, constructing an expert rule base, using the rule base to reason and score each image region, and summarizing to obtain a suitability score map of the region.

[0038] In another optional implementation, in step S100, identifying and screening candidate substation locations that meet the preset construction conditions may further include extracting suitable land types for industrial and mining storage land and public facility land from land use data to form an initial planar layer, establishing buffer zones (500 meters) for prohibited or restricted areas of ecological red lines, geological disaster-prone areas, and cultural heritage protection areas, and performing overlay analysis on the initial suitable layer and the buffer zones of various restricted areas to eliminate restricted areas, merge and screen the remaining areas, and the centroid or the center of the largest inscribed circle can be used as a candidate site.

[0039] In this embodiment of the application, step S200, the formation of the station line scheme includes steps S201~S202: S201: Design a unified individual coding scheme, treating the complete planning scheme as an optimization entity. The data structure includes the substation coordinates selected from the candidate set and the ordered grid node paths connecting the power source point to the current station.

[0040] S202: The coding structure binds the two strongly coupled decision variables of station selection and line drawing into the same entity, enabling the algorithm to coordinate the adjustment of the two within a unified framework, effectively avoiding the error accumulation and local optimum problems caused by phased optimization.

[0041] In an optional implementation, in step S200, the formation of the station route scheme further includes converting the grid space model into a graph, where each grid is a node, the connection between adjacent grids is an edge, the weight of the edge is the passage cost, and the encoding of an individual is divided into substation node number and path decision sequence; the path decision sequence is a direction instruction or the selection of an edge. During decoding, starting from the starting point, the path moves in the graph according to the instruction sequence, and the node finally reached is the substation location, and the sequence of nodes traversed is the path.

[0042] In another optional implementation, in step S200, the formation of the station line scheme may further include encoding an individual including substation coordinates and a set of path control point coordinates. During decoding, a spline curve or polynomial curve is used to sequentially connect the starting point, all control points, and the substation end point to generate a continuous and smooth curve. The current curve is then rasterized and mapped back to the raster space for evaluation.

[0043] Furthermore, in step S200, the establishment guidance information includes steps S211~S212: S211: Dynamic guidance information is constructed for the update of each path node based on the wind farm guidance mechanism; The dynamic guidance information is composed of two weighted vectors: one part is the negative gradient calculated based on the passage cost map, which is approximated by the central difference method and points to the place where the cost decreases the fastest, so as to achieve local risk avoidance; the other part is the unit direction vector from the current node to the target substation of the path, which is used for global benefit. The formula for constructing dynamic guidance information is expressed as follows: in, This is the guidance information for the current node. For local gradient information, The direction information is the unit direction vector pointing from the current node to the target substation along the path. As the guiding weighting factor, c and d are the row and column indices of the raster cell where the current path node is located. Let (c+1, d) be the passage cost. Let (c-1, d) be the passage value. Let (c, d+1) be the passage cost. Let (c, d-1) be the passage value. Let be the side length of each grid cell in the grid space model. It is approximately equal to.

[0044] S212: By adjusting the guiding weight factor to balance local obstacle avoidance and global trend, the mechanism transforms static geographical cost information into dynamic search driving force, enabling path nodes to simultaneously take into account local terrain constraints and global goal orientation when updating.

[0045] Furthermore, in this embodiment of the application, in step S200, calculating the fitness value of the optimization target includes steps S221~S222: S221: During calculation, the path length is calculated as the total geometric length of all path segments. The ecological impact is obtained by multiplying the length of the path crossing each grid with the ecological sensitivity score. The construction cost is estimated using cost factors in combination with the path slope and length. The power supply reliability is quantified by the connection strength between the substation and the main power grid. The formula for calculating ecological impact is expressed as follows: in, The total ecological impact assessment value of the individual i-line planning scheme is given. For each individual grid cell traversed by the path, Let h be the length of the line segment within the grid cell h. To score the ecological sensitivity of grid cell h, This refers to all the raster cells traversed by the path; The formula for calculating construction costs is expressed as follows: in, This represents the estimated total construction cost for individual station and line planning scheme i. is the construction cost factor per unit length of the grid cell h.

[0046] S222: The min-max normalization method is used to map each original target value to a unified interval, eliminating the influence of dimensions and obtaining normalized fitness values, laying the foundation for multi-target comparison and ranking.

[0047] In an optional implementation, in step S200, the calculation of the fitness value of the optimization target further includes examining the cost value of all grids within a 5x5 window, taking the grid where the current node is located as the center, finding the grid with the lowest cost value in the neighborhood, and calculating the direction vector from the current node to the center of the grid with the lowest cost, using the current direction as local guidance information.

[0048] In another optional implementation, in step S200, the calculation of the fitness value of the optimization target may further include setting the target substation location as the lowest point of the attraction potential, solving the Poisson equation or using the fast travel method in the entire grid space to calculate the global situation field, where the potential value at any point in the space represents the cumulative cost from the current point to the target point, and for path nodes, the local guidance information is the negative gradient of the global situation field at the current point.

[0049] In step S300, the iterative update of the population includes using an improved Golden Kite optimization algorithm to iteratively evolve the population. After generating the initial population, fitness evaluation, elite memory pool update, population update, and perturbation repair operations are performed in each generation, specifically including steps S301 to S305: S301: By establishing an elite memory pool to retain excellent non-dominated solutions from previous generations, after each iteration, the excellent non-dominated solutions are merged with the current population, and non-dominated sorting and crowding calculation are performed. Solutions in dense regions are eliminated first to maintain frontier diversity. The formula for calculating the crowding degree of the new solution and the solutions in the pool is expressed as: in, Let be the crowding level of the i-th individual. Let i+1 be the normalized fitness value of the i+1th individual on the kth optimization objective dimension. Let be the normalized fitness value of the (i-1)th individual on the k-th optimization objective dimension. This represents the maximum fitness value in the k-th objective dimension among all current individuals. It represents the minimum fitness value in the k-th target dimension among all current individuals. To optimize the total number of objectives, To optimize the target index, For individual indexes.

[0050] S302: During the population update phase, the geographic gradient guidance information calculated from the individual location, the Gaussian random perturbation, and the guidance vector that integrates elite individual information are adjusted in a coordinated manner, and random noise is applied to the path nodes using the local perturbation operator to enhance diversity. The formula for individual renewal in an elite population is expressed as: in, Let i be the current position of the i-th individual in the t-th iteration. Let i be the new position of the i-th individual after the (t+1)-th iteration. The wind field propulsion coefficient, This is the wind field guiding vector. This represents the local disturbance intensity coefficient. It is a random perturbation vector.

[0051] S303: The local perturbation strategy includes, after each iteration of population position update, traversing the path node sequence of each individual with a preset probability, superimposing a Gaussian distributed random offset with zero mean and controllable variance on the grid coordinates of the selected node, and after perturbation update, checking whether all path nodes of each individual fall into the infeasible grid, and for any node that falls into the infeasible grid, adjusting the coordinates to the center of the nearest feasible grid within the surrounding eight neighborhoods and with a passage cost lower than the preset feasibility threshold; The local perturbation operator formula is expressed as: in, Let l be the position of the l-th node in the path during the t-th iteration. This represents the new position of the current path node after perturbation in the (t+1)th iteration. As a disturbance factor, To have a mean of 0 and a variance of A random variable with a Gaussian distribution.

[0052] S304: Iterative updates to the population include, for each individual in the population, using the composite vector as the main guiding direction, the introduced elite guiding vector as the auxiliary direction, and the introduced random offset as the exploration term, calculating the incremental change of position through a weighted summation formula, and adding the increment to the current position to generate the position of a new generation of individuals; The elite guiding vector is represented as: in, For individual location The wind field steering vector when the path node is located at grid coordinates (a, b). This represents the negative direction of the grid cost gradient vector at the current position. The target guiding coefficient, This represents the direction vector of the elite individual selected from the elite memory pool; The formula for adjusting the update direction under the guidance of elites is expressed as follows: in, This is the adjusted composite direction vector used to update ordinary individuals. For balance coefficient, This represents the current position of the j-th individual that is currently being updated.

[0053] S305: The algorithm iterates until the convergence condition of the maximum number of iterations or the frontier improvement amount being lower than the threshold is met.

[0054] Furthermore, in step S300, the output Pareto front solution set includes steps S311~S313: S311: After the optimization process converges, the algorithm outputs the first-level individuals obtained by non-dominated sorting in the final generation population, and constructs a Pareto front solution set that is non-dominated on the four objectives of path length, ecological impact, construction cost and power supply reliability.

[0055] S312: Based on the actual preferences of the project, the planning engineer may prioritize ecological protection or pursue the lowest cost, and flexibly select the optimal solution from the current solution set.

[0056] S313: The final selected solution is visualized with high precision through a geographic information system platform, directly serving the engineering feasibility study report and preliminary design work, realizing a closed loop from data analysis to final decision-making.

[0057] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0058] Example 3, referring to Figure 2This is the third embodiment of the present invention, which provides an intelligent site selection and route selection system for power transmission and transformation projects, including a spatial data fusion and cost modeling module, a candidate site intelligent screening module, a problem unified coding and modeling module, a geographic gradient guidance and path search module, and a multi-objective collaborative optimization and decision output module.

[0059] The spatial data fusion and cost modeling module is used to integrate multi-source heterogeneous geospatial data from remote sensing, topography, land use and ecological protection areas. It constructs a digital spatial base through unified coordinates and rasterization processing, and uses a preset cost function to calculate the cost value of each raster unit, which comprehensively reflects ecological sensitivity and engineering economic costs, to generate a cost map.

[0060] The intelligent candidate site screening module is used to automatically analyze remote sensing images using deep learning image segmentation technology, identify and extract areas with construction conditions, generate a site suitability probability map, and combine multi-dimensional engineering evaluation indicators such as load center distance, geological stability, and traffic accessibility to comprehensively quantify and rank candidate areas, and output a set of candidate substation locations.

[0061] The unified coding and modeling module is used to encode the spatial coordinates of the substation and the sequence of transmission line path nodes connecting the current station in an integrated manner to form a complete solution expression, and to adjust the station location and path direction simultaneously.

[0062] The geographic gradient guidance and path search module is used to dynamically calculate the guidance vector for each node to be adjusted on the path during the optimization iteration process, transforming geographic environmental information into a search driving force with physical meaning.

[0063] The multi-objective collaborative optimization and decision output module is used to construct a multi-objective fitness function evaluation scheme that includes path length, ecological impact, cost and reliability, adopt an elite retention strategy to maintain high-quality solutions, combine non-dominated sorting and crowding calculation to maintain the diversity and distribution of solutions, introduce local perturbation and feasibility repair mechanisms to handle spatial hard constraints, and output a Pareto optimal solution set that achieves the best trade-off on multiple objectives.

[0064] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0065] Example 4, the fourth embodiment of the present invention, differs from the previous three embodiments in that: If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0066] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0067] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0068] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

Claims

1. A method for intelligent site selection and route selection in power transmission and transformation projects, characterized in that: include, Multi-source spatial data is acquired and fused to construct a unified grid spatial model. Based on the grid spatial model, the passage cost value is calculated for each grid cell to generate a passage cost map. Based on the spatial data, candidate substation locations that meet the preset construction conditions are identified and screened. The candidate locations are comprehensively scored according to the evaluation indicators, and the set of candidate locations is determined based on the scoring results. The selected substation locations and connection paths are uniformly coded to form a station-line scheme. In the optimization iteration, the path nodes of each individual are updated to construct guidance information, and the fitness value of the optimization target is calculated for each individual. The target values ​​are then normalized. With the goal of optimizing multi-objective fitness, the population is iteratively updated. An elite solution set that stores non-dominated solutions is introduced to guide the direction of population evolution. A local perturbation strategy is combined until the convergence condition is met. When the termination condition is met, the Pareto front solution set is output.

2. The intelligent site selection and route selection method for power transmission and transformation projects as described in claim 1, characterized in that: The process of generating the access cost map involves integrating geographic information data within the study area and converting it into a uniform raster format, establishing an integrated spatial representation model, calculating the access cost value for each grid cell in the rasterized space, and determining the final value by setting different weight coefficients for the cost, thereby generating the cost map.

3. The intelligent site selection and route selection method for power transmission and transformation projects as described in claim 2, characterized in that: The process of determining the candidate location set based on the scoring results includes: automatically analyzing and processing image information from multi-source spatial data, extracting a probability distribution map representing the suitability of construction in a geographical area, and preliminarily delineating highly suitable areas based on the probability distribution map by setting a threshold. For each initially identified region, a multi-indicator quantitative evaluation is conducted, and the comprehensive score of each region is calculated based on a weighted scoring function; Based on the overall score, a predetermined number of regional center points with the highest ranking are selected to form a set of candidate substation locations for entering the optimization process.

4. The intelligent site selection and route selection method for power transmission and transformation projects as described in claim 3, characterized in that: The construction guidance information includes querying the passage cost map for the nodes to be updated in the path sequence, calculating the rate of change of cost between the current node's grid and its adjacent grids, and determining local gradient information indicating the direction of cost reduction. Calculate the unit direction vector from the current node to the end substation location of its path, as the direction information, and synthesize the local gradient information and the direction information into a vector according to the preset weight coefficients to generate a composite vector, which serves as the guiding information for the current node to move in this iteration.

5. The intelligent site selection and route selection method for power transmission and transformation projects as described in claim 4, characterized in that: The fitness values ​​of the optimization objectives include the path length objective, the ecological impact objective, the construction cost objective, and the power supply reliability objective. The original values ​​of the four objectives are normalized and converted into dimensionless fitness scores. The objective of path length calculation includes calculating the total extension length of the path encoded by an individual across all raster cells; The calculation of ecological impact targets includes identifying the ecological sensitivity score of each grid cell traversed by the path, multiplying the length of the path segment in each grid cell by the ecological sensitivity score of the current grid cell, and summing the results to obtain the ecological impact weighted total length. The calculation of construction cost targets includes calculating the slope of the area traversed by the path based on digital elevation model data, and combining the path length to calculate the estimated total construction cost through a comprehensive cost function. The calculation of power supply reliability targets includes assessing the topological connection distance between the end-point substation location and key nodes in the existing main power grid, and quantifying the level of power supply reliability.

6. The intelligent site selection and route selection method for power transmission and transformation projects as described in claim 5, characterized in that: The guidance of population evolution includes performing a non-dominated sort on all current individuals after each generation iteration, and storing all non-dominated solutions of the first level of sorting into a fixed-size storage structure in the elite memory pool. The update rules for the elite memory pool include: when a newly generated solution dominates an old solution in the pool, the new solution replaces the old solution; when the new solution does not dominate any of the solutions in the pool, the crowding degree between the new solution and the solutions in the pool is calculated; when the target space region where the new solution is located is sparser than the solutions in the pool, the new solution replaces the solution with the highest current crowding degree. The formula for calculating the crowding degree of the new solution and the solutions in the pool is expressed as: in, Let be the crowding level of the i-th individual. Let i+1 be the normalized fitness value of the i+1th individual on the kth optimization objective dimension. Let be the normalized fitness value of the (i-1)th individual on the k-th optimization objective dimension. This represents the maximum fitness value in the k-th objective dimension among all current individuals. It represents the minimum fitness value in the k-th target dimension among all current individuals. To optimize the total number of objectives, To optimize the target index, For individual indexes; The formula for individual renewal in an elite population is expressed as: in, Let i be the current position of the i-th individual in the t-th iteration. Let i be the new position of the i-th individual after the (t+1)-th iteration. The wind field propulsion coefficient, This is the wind field guiding vector. This represents the local disturbance intensity coefficient. It is a random perturbation vector; The local perturbation strategy includes, after each iteration of population position update, traversing the path node sequence of each individual with a preset probability, superimposing a Gaussian distributed random offset with zero mean and controllable variance on the grid coordinates of the selected nodes, and after the perturbation update, checking whether all path nodes of each individual fall into infeasible grids. For any node that falls into an infeasible grid, adjusting the coordinates to the center of the nearest feasible grid within the surrounding eight neighborhoods and with a passage cost lower than a preset feasibility threshold. The local perturbation operator formula is expressed as: in, Let l be the position of the l-th node in the path during the t-th iteration. This represents the new position of the current path node after perturbation in the (t+1)th iteration. As a disturbance factor, To have a mean of zero and a variance of A random variable with a Gaussian distribution.

7. The intelligent site selection and route selection method for power transmission and transformation projects as described in claim 6, characterized in that: The iterative update of the population includes, for each individual in the population, using the composite vector as the main guiding direction, the introduced elite guiding vector as the auxiliary direction, and the introduced random offset as the exploration term, calculating the incremental change of position through a weighted summation formula, and superimposing the increment onto the current position to generate the position of a new generation of individuals; The elite guiding vector is represented as: in, For individual location The wind field steering vector when the path node is located at grid coordinates (a, b). This represents the negative direction of the grid cost gradient vector at the current position. The target guiding coefficient, This represents the direction vector of the elite individual selected from the elite memory pool; The formula for adjusting the update direction under the guidance of elites is expressed as follows: in, This is the adjusted composite direction vector used to update ordinary individuals. For balance coefficient, This represents the current position of the j-th individual that is currently being updated; The convergence conditions include reaching the preset maximum number of iterations, outputting the final solution set, and extracting all individuals within the first non-dominated level after performing non-dominated sorting.

8. A smart site selection and route selection system for power transmission and transformation projects, employing the smart site selection and route selection method for power transmission and transformation projects as described in any one of claims 1 to 7, characterized in that, It includes a spatial data fusion and cost modeling module, a candidate site intelligent screening module, a problem unified coding and modeling module, a geographic gradient guidance and path search module, and a multi-objective collaborative optimization and decision output module; The spatial data fusion and cost modeling module is used to integrate multi-source heterogeneous geospatial data from remote sensing, topography, land use and ecological protection areas. It constructs a digital spatial base through unified coordinates and rasterization processing, and uses a preset cost function to calculate the cost value of each raster unit, which comprehensively reflects ecological sensitivity and engineering economic costs, to generate a cost map. The candidate site intelligent screening module is used to automatically analyze remote sensing images using deep learning image segmentation technology, identify and extract areas with construction conditions, generate a site suitability probability map, and combine multi-dimensional engineering evaluation indicators such as load center distance, geological stability, and traffic accessibility to comprehensively quantify and rank the candidate areas, and output a set of candidate substation locations. The unified coding and modeling module is used to encode the spatial coordinates of the substation and the sequence of transmission line path nodes connecting the current station in an integrated manner to form a complete solution expression, and to adjust the station location and path direction simultaneously. The geographic gradient guidance and path search module is used to dynamically calculate the guidance vector for each node to be adjusted on the path during the optimization iteration process, and to transform the geographic environment information into a search driving force with physical meaning. The multi-objective collaborative optimization and decision output module is used to construct a multi-objective fitness function evaluation scheme that includes path length, ecological impact, cost and reliability, adopt an elite retention strategy to maintain high-quality solutions, combine non-dominated sorting and crowding calculation to maintain the diversity and distribution of solutions, introduce local perturbation and feasibility repair mechanisms to handle spatial hard constraints, and output a Pareto optimal solution set that achieves the best trade-off on multiple objectives.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent site selection and route selection method for power transmission and transformation projects as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent site selection and route selection method for power transmission and transformation projects as described in any one of claims 1 to 7.

Images