Intelligent route selection method and system for overhead power transmission line

Abstract

This invention proposes an intelligent route selection method and system for overhead transmission lines, belonging to the field of transmission line route selection technology. The method includes: extracting terrain and environmental data of the planned transmission line area; generating a feasible space for planning parameters; solving for the optimal feasible solution using gradient descent within the feasible space; using the optimal feasible solution as the inner approximation set and the feasible space as the outer approximation set; solving an optimization problem in the outer approximation set to maximize the minimum distance to the inner approximation set; determining candidate planning parameter points; inputting these points into a line performance score prediction model; pre-screening and validating the planning parameters based on the predicted performance scores to update the outer and inner approximation sets; repeating the iteration until convergence; clustering the converged inner approximation set to generate the final planning parameter set; and generating the final three-dimensional overhead transmission line using a path generator. This invention improves the automation level, efficiency, and accuracy of transmission line route selection.

Description

Technical Field

[0001] This invention belongs to the field of power transmission line selection technology, and more specifically, relates to an intelligent method and system for selecting overhead power transmission lines. Background Technology

[0002] Traditional route selection relies heavily on the engineering experience of designers, typically involving manual interpretation and drawing on two-dimensional maps or simplified three-dimensional terrain. This method is time-consuming, subjective, and struggles to achieve systematic and global optimization in complex terrain environments with multiple constraints, easily leading to unreasonable routes or excessive costs.

[0003] Many studies employ a single optimization algorithm, such as, but not limited to, A* algorithm, genetic algorithm, and particle swarm optimization, to automatically search for paths under a fixed cost model. These methods have significant limitations: First, in complex terrain, high-precision engineering cost calculations, such as, but not limited to, accurate tower foundation costs, are extremely time-consuming, severely restricting the search depth and breadth of the optimization algorithm and making it difficult to find a truly high-quality global solution within a limited time.

[0004] Traditional optimization methods lack a mechanism for exploring the near-optimal solution space, which cannot guarantee the quality of the obtained solution or quantify the gap between the current solution and the global optimum. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent route selection method and system for overhead transmission lines. By integrating three-dimensional parametric modeling, lightweight machine learning model evaluation, and a mathematically provably convergent near-optimal space exploration framework, the automation level, global optimization capability, efficiency, and route selection accuracy of the transmission line route selection process are improved. This reduces excessive reliance on human experience, high computational time costs, and engineering decision-making risks caused by poor quality transmission line route selection schemes.

[0006] The present invention adopts the following technical solution.

[0007] The first aspect of the present invention provides an intelligent route selection method for overhead transmission lines, comprising: Obtain topographic environmental data and transmission line planning parameters of the planned area. Within the feasible space of the planning parameters, with the goal of minimizing the total cost of the line path, solve for the optimal feasible solution. Use the optimal feasible solution as the inner approximation set and the feasible space as the outer approximation set. A line performance score prediction model is constructed based on deep neural networks; Using the minimum distance between the outer approximation set and the inner approximation set as the planning index, candidate planning parameters are searched in the outer approximation set to maximize the planning index. The candidate planning parameters are input into the line performance score prediction model to obtain the predicted performance score. Candidate planning parameters are pre-screened based on the predicted performance score. The pre-screened candidate planning parameters are verified based on the total cost of the line path. The verified candidate planning parameters are moved from the outer approximation set to the inner approximation set. Based on the latest outer approximation set and the latest inner approximation set, the pre-screening and verification are repeated until convergence, and the converged inner approximation set is determined. Cluster the converged inner approximate set, select the planning parameter closest to the cluster center from each cluster to generate the final planning parameter set, and generate the final three-dimensional overhead transmission line through the path generator.

[0008] Preferably, the feasible space for generating planning parameters includes: Transmission lines are represented by B-spline curves containing multiple control points. The starting and ending points of the transmission lines are fixed as the first and last control points of the B-spline curve. The horizontal and vertical offsets of the remaining control points relative to the line connecting the first and last control points are set as planning parameters. Obtain topographic environmental data of the transmission line planning area, use the topographic environmental data to construct minimum clearance constraints, maximum slope constraints and obstacle avoidance distance constraints, determine the set of all planning parameters that satisfy the constraints, and generate the feasible space of the planning parameters.

[0009] Preferably, using the optimal feasible solution as the inner approximation set and the feasible space as the outer approximation set includes: The path generator uses planning parameters in the feasible space and the starting and ending points of the transmission line to generate a three-dimensional path, and repeatedly samples the three-dimensional path to obtain multiple coordinate sequences. Calculate the total cost of the route corresponding to each coordinate sequence; The planning parameters corresponding to the minimum total cost of the transmission route are taken as the optimal feasible solution. The optimal feasible solution and the starting and ending points of the transmission line constitute an inner approximate set. The planning parameters other than the optimal feasible solution in the feasible space and the starting and ending points of the transmission line constitute an outer approximate set.

[0010] Preferably, historical data of the transmission line is acquired, including terrain and environmental data and planning parameters of the historical transmission line tasks; The historical data of transmission lines and their corresponding total route costs are standardized and used as the sample set. The standardized total route costs are then negative and set as the true labels. Based on the sample set and the real labels, a deep neural network is trained by minimizing the mean square error between the predicted line performance score and the real label to obtain the line performance score prediction model.

[0011] Preferably, the minimum distance from the outer approximation set to the inner approximation set is used as the planning index. Candidate planning parameters are searched in the outer approximation set to maximize the planning index, as expressed by the following formula:

[0012] In the formula, Indicates the candidate planning parameters. express Maximize the planning index at the location, Describing the outer approximation set Planning parameters to inner approximate set Planning parameters The minimum distance is the planning parameter. Planning indicators Represents Euclidean distance. This means that all constraints are satisfied. of The goal is to find the planning parameters that maximize the planning indicators as candidate planning parameters.

[0013] Preferably, the candidate planning parameters pre-screened based on the prediction performance score include: The candidate planning parameters are input into the line performance score prediction model, the predicted performance score is output, and the performance threshold is determined based on the maximization planning index of the candidate planning parameters. The predicted performance score is compared with a performance threshold to determine the candidate planning parameters that pass the pre-screening, including: If the predicted performance score is not less than the performance threshold, the candidate planning parameters are deemed to have passed the pre-screening. If the predicted performance score is less than the performance threshold, the candidate planning parameters are determined to have failed the pre-screening. Based on the set relaxation factor and the candidate planning parameters that have passed the pre-screening, new candidate planning parameters are constructed to replace the candidate planning parameters that have failed the pre-screening, thereby updating the outer approximation set. In the latest outer approximation set, the candidate planning parameters are re-determined to maximize the planning index.

[0014] Preferably, the performance threshold is determined based on the planning index of the candidate planning parameters, expressed by the following formula:

[0015] In the formula, Indicates the performance threshold. Indicates the baseline performance threshold. Indicates the exploration coefficient. It represents the maximization of planning index based on the candidate planning parameters, which are the inner approximation set with the optimal feasible solution and the outer approximation set with the feasible space.

[0016] Preferably, determining the converged inner approximation set includes: Based on the candidate planning parameters selected through pre-screening, and the starting and ending points of the transmission lines to form candidate paths, the total route cost of the candidate paths is calculated. If the total cost of the candidate path is not greater than the set cost threshold, the candidate planning parameter is deemed to have been successfully verified, and the candidate planning parameter is moved from the outer approximation set to the inner approximation set. If the total cost of a candidate path exceeds the set cost threshold, the candidate planning parameter verification is deemed to have failed. The external approximation set is then updated based on the total cost of the candidate path and the planning indicators of the failed candidate planning parameters. The latest candidate planning parameters and planning indices at the latest candidate planning parameters are determined based on the latest outer approximation set and the latest inner approximation set. If the planning indices at the candidate planning parameters are greater than the set convergence threshold, it is determined that the convergence has not been achieved. The latest candidate planning parameters are then pre-screened and verified repeatedly, and the next iteration begins until the converged inner approximation set is obtained.

[0017] Preferably, the external approximation set is updated based on the total cost of the route path and the planning indicators of the candidate planning parameters that failed verification, as expressed by the following formula:

[0018] In the formula, Represents the updated outer approximation set. express Planning parameters in Let the outer approximate set be updated in the r-th iteration. This represents the total cost of the route path corresponding to the candidate planning parameters that failed validation. Indicates the exploration coefficient. Planning parameters indicating failed validation Planning indicators Representing the inner approximate set Planning parameters in Indicates the total cost of the route path. The gradient vector at that point, This represents the transpose operator. This represents the theoretically optimal cost. Indicates cost margin.

[0019] A second aspect of the present invention provides an intelligent route selection system for overhead transmission lines, which operates the intelligent route selection method for overhead transmission lines described in the first aspect, comprising: The inner and outer approximation set generation module is used to obtain the terrain environment data of the transmission line planning area and the planning parameters of the transmission line. In the feasible space of the planning parameters, the optimal feasible solution is solved with the goal of minimizing the total cost of the line path. The optimal feasible solution is used as the inner approximation set, and the feasible space is used as the outer approximation set. The model building module is used to build a line performance score prediction model based on deep neural networks. The solution module uses the minimum distance between the outer approximation set and the inner approximation set as the planning index. It searches for candidate planning parameters in the outer approximation set that maximize the planning index, inputs the candidate planning parameters into the line performance score prediction model to obtain the predicted performance score, pre-screens candidate planning parameters based on the predicted performance score, and verifies the pre-screened candidate planning parameters based on the total cost of the line path. The verified candidate planning parameters are moved from the outer approximation set to the inner approximation set. Based on the latest outer approximation set and the latest inner approximation set, the pre-screening and verification are repeated until convergence, and the converged inner approximation set is determined. The route generation module is used to cluster the converged inner approximate set, select the planning parameters closest to the cluster center from each cluster to generate the final planning parameter set, and generate the final three-dimensional overhead transmission line through the path generator.

[0020] Compared with the prior art, the beneficial effects of the present invention include at least the following: This invention solves the technical problem of low search efficiency of optimization algorithms caused by the extremely time-consuming high-precision engineering cost simulation calculation in optimization search by training a lightweight machine learning line performance prediction score (deep neural network) using historical engineering data. It improves the real-time performance evaluation and selection of transmission line schemes, and reduces the time cost and computational resource consumption of the transmission line selection optimization process. It is suitable for generating transmission line schemes for long distances and complex terrain. This invention combines the pre-screening mechanism of the line performance score prediction model with the mathematically rigorous ORACLE near-optimal space exploration iterative framework, which solves the technical problems of traditional optimization algorithms being prone to getting trapped in local optima and the inability to guarantee the quality of solutions. By maximizing minimum distance sampling and shrinking the outer approximation set, it achieves a mathematically rigorous exploration of the high-quality solution space, rather than a random search, thereby improving the global near-optimality and reliability of the final transmission line scheme set and reducing the engineering decision-making risks and later change costs caused by poor scheme quality. This invention extracts the most representative scheme from a large number of verified high-quality transmission line route selection schemes by clustering the converged inner approximate set, thereby reducing the blindness and coordination costs in the final decision-making stage of transmission line route selection. Attached Figure Description

[0021] Figure 1This is a schematic diagram of the intelligent overhead transmission line selection process provided in accordance with an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0023] Embodiment 1 of the present invention provides an intelligent line selection method for overhead transmission lines, such as... Figure 1 As shown, it includes the following steps: Step 1: Obtain the topographic environment data of the transmission line planning area and the planning parameters of the transmission line. In the feasible space of the planning parameters, with the goal of minimizing the total cost of the line path, solve for the optimal feasible solution. Use the optimal feasible solution as the inner approximation set and the feasible space as the outer approximation set.

[0024] Specifically, step 1 includes: Step 1.1: Obtain the terrain and environmental data of the current power transmission line planning area.

[0025] Topographic environmental data includes, but is not limited to: Euclidean distance between the starting and ending points of the transmission line, average slope, topographic relief, and obstacle density.

[0026] Step 1.2: The transmission line is represented by a B-spline curve containing multiple control points. The starting point and ending point of the transmission line are fixed as the first and last control points of the B-spline curve. The horizontal and vertical offsets of the remaining control points relative to the line connecting the first and last control points are set as planning parameters.

[0027] In Example 1, the remaining control points refer to a set of movable points located between the start and end points, used to control and shape the overall curve shape, when representing the transmission line path using a B-spline curve (or other parametric curves). These points are separate from the fixed start and end points. The positions of the remaining control points are adjustable; within the constraints of minimum clearance, maximum gradient, and obstacle avoidance distance, the feasible space for planning parameters is determined by adjusting the positions of the remaining control points.

[0028] Step 1.3: Obtain the topographic environment data of the transmission line planning area, use the topographic environment data to construct minimum clearance constraints, maximum slope constraints and obstacle avoidance distance constraints, determine the set of all planning parameters that meet the constraints, and generate the feasible space of the planning parameters.

[0029] Step 1.4: Solve for the optimal feasible solution in the feasible space of the planning parameters with the goal of minimizing the total cost of the route path. Use the optimal feasible solution as the inner approximation set and the feasible space as the outer approximation set.

[0030] Furthermore, step 1.4 includes: Step 1.4.1: Using the B-spline Generator, a three-dimensional path is generated by utilizing the planning parameters in the feasible space and the starting and ending points of the transmission line. The three-dimensional path is then repeatedly sampled to obtain multiple coordinate sequences.

[0031] The path generator is invoked to convert the abstract planning parameters into a three-dimensional path that can be used for geometric calculation. This process generates a continuous three-dimensional path curve based on the B-spline curve model and performs dense sampling on it to obtain a discrete point coordinate sequence.

[0032] Step 1.4.2: Calculate the total cost of the route corresponding to each coordinate sequence; the total cost of the route corresponding to the planning parameters calculated in the first iteration is set as the initial historical optimal cost. The total cost of the route includes the cost of the wiring material and the total cost of the towers and foundations.

[0033] The total cost of the route is expressed by the following formula: (1) In the formula, This represents the total cost of the route corresponding to the current planning parameters. The cost of the wire is expressed by the following formula: (2) In the formula, This represents the comprehensive unit price per unit length of wire. When representing the optimal feasible solution. Represents the optimal feasible solution The corresponding wire cost, The total length of the 3D path generated by the planning parameters is expressed by the following formula: (3) In the formula, Indicates the sampling point. This represents the total number of sampling points. To calculate the length of the continuous path, discrete sampling points are taken at equal intervals along the path curve. The B-spline Generator generates a continuous three-dimensional path curve based on the planning parameters. Indicates the first The coordinates of each sampling point in the eastward direction of the three-dimensional path. Indicates the first The coordinates of each sampling point in the north direction of the three-dimensional path. Indicates the first The ground elevation of each sampling point along the three-dimensional path Indicates the first The three-dimensional coordinates of the nth sampling point, i.e., the nth The coordinates of the next adjacent point of each sampling point.

[0034] The total cost of the tower and foundation is expressed by the following formula: (4) In the formula, This indicates the total number of towers required for the current path. Indicates the first The type of tower at each tower location, This indicates the cost of the tower body, determined based on the type of tower. This indicates the unit cost of adjusting the foundation slope. This indicates the unit cost of adjusting the foundation bearing capacity. Indicates the first The ground slope of each tower site, Indicates the first The bearing capacity of the foundation of each tower site.

[0035] Step 1.4.3: The planning parameters corresponding to the minimum total cost of the transmission route are taken as the optimal feasible solution. The optimal feasible solution and the starting point and ending point of the transmission line constitute an inner approximate set. The planning parameters other than the optimal feasible solution in the feasible space and the starting point and ending point of the transmission line constitute an outer approximate set.

[0036] In Example 1, the route selection problem is mathematically formalized by extracting terrain features in step 1.1 and defining a parameterized model in step 1.2. Then, in step 1.3, a high-quality optimal feasible solution (optimal planned route) is quickly located and a range of optimal feasible solutions is defined through optimization methods such as gradient descent. Finally, this benchmark solution and range are used to initialize the starting point for subsequent system exploration, providing a clear objective and efficient starting direction for the entire intelligent route selection algorithm. This transforms route selection from an empirical search to a quantitative optimization problem based on data and parameters, improving the automation and quantification level of the transmission line route selection process and reducing the subjective dependence of scheme design on human experience and the complexity of transmission line modeling.

[0037] Step 2: Construct a line performance score prediction model based on a deep neural network. The deep neural network in this invention is a multilayer perceptron.

[0038] Specifically, step 2 includes: Step 2.1: Standardize the historical data of transmission lines and their corresponding total route costs to form a sample set, and set the negative of the standardized total route costs as the true label.

[0039] Obtain historical data on transmission lines, including terrain and environmental data and planning parameters for historical transmission line tasks.

[0040] Historical scenarios for power transmission lines include mountainous wind power transmission lines, main power grid lines in plains areas, and lines crossing rivers. Topographic data for mountainous wind power transmission lines includes a length of 15km, an average slope of 8%, high undulation, and low obstacle density (mainly mountains). Planning parameters include path offsets for bypassing mountain ridges to connect wind farms and substations. Topographic data for main power grid lines in plains areas includes a length of 30km, an average slope of 1%, low undulation, and medium obstacle density (including farmland, villages, and roads). Planning parameters include path offsets for crossing roads multiple times and avoiding densely populated villages. Topographic data for lines crossing rivers includes a length of 8km, an average slope of 0.5%, low undulation, and low obstacle density (large bodies of water, waterways). Planning parameters include path offsets for using high towers and directly crossing rivers.

[0041] The total cost of the transmission line path in different historical scenarios in the historical data of the transmission line is standardized by Z-score, and then the negative value is set as the true label.

[0042] Step 2.2: Based on the sample set and real labels, a deep neural network is trained by minimizing the mean square error between the predicted line performance score and the real label to obtain the line performance score prediction model, which is used to learn the mapping relationship from the terrain environment data and planning parameters of the historical scene of the transmission line to the predicted line performance.

[0043] This invention performs Z-score standardization on the total cost of transmission line paths for different historical tasks in historical data, then negates the Z-score. The standardized performance score prediction model predicts the negative standardized value, transforming the cost minimization problem into a score maximization problem that is more in line with machine learning conventions. The negativeized standardized total cost of the transmission line path generates the true negative normalized cost of the line, which is set as the true label for the deep neural network.

[0044] Deep neural networks use the mean squared error between the predicted line performance score and the true label as the loss function, and are trained by minimizing the loss function.

[0045] Step 3: Using the minimum distance from the outer approximation set to the inner approximation set as the planning index, candidate planning parameters are searched in the outer approximation set to maximize the planning index. The candidate planning parameters are input into the line performance score prediction model to obtain the predicted performance score. Candidate planning parameters are pre-screened based on the predicted performance score. The pre-screened candidate planning parameters are verified based on the total cost of the line path. The verified candidate planning parameters are moved from the outer approximation set to the inner approximation set. Based on the latest outer approximation set and the latest inner approximation set, the pre-screening and verification are repeated until convergence, and the converged inner approximation set is determined.

[0046] Specifically, step 3 includes: Step 3.1: Using the minimum distance from the outer approximation set to the inner approximation set as the planning index, find candidate planning parameters in the outer approximation set that maximize the planning index, expressed by the following formula: (5) In the formula, Indicates the candidate planning parameters. express Maximize the planning index at the location, Describing the outer approximation set Planning parameters to inner approximate set Planning parameters The minimum distance is the planning parameter. Planning indicators Represents Euclidean distance. This means that all constraints are satisfied. of The goal is to find the planning parameters that maximize the planning indicators as candidate planning parameters.

[0047] Step 3.2: Input the candidate planning parameters into the line performance score prediction model, output the predicted performance score, and determine the performance threshold based on the maximization planning index of the candidate planning parameters, expressed by the following formula: (6) In the formula, Indicates the performance threshold. Indicates the baseline performance threshold. , Represents the normalization function. This represents the theoretically optimal cost. Indicates cost margin, Indicates the exploration coefficient. It represents the maximization of planning index based on the candidate planning parameters, which are the inner approximation set with the optimal feasible solution and the outer approximation set with the feasible space.

[0048] Step 3.3 compares the predicted performance score with the performance threshold to determine the candidate planning parameters that pass the pre-screening, including: If the predicted performance score is not less than the performance threshold, the candidate planning parameters are deemed to have passed the pre-screening. If the predicted performance score is less than the performance threshold, the candidate planning parameters are determined to have failed the pre-screening. Based on the set relaxation factor and the candidate planning parameters that have passed the pre-screening, new candidate planning parameters are constructed to replace the candidate planning parameters that have failed the pre-screening, thereby updating the outer approximation set. In the latest outer approximation set, the candidate planning parameters are re-determined to maximize the planning index.

[0049] In Example 1, by introducing a line performance score prediction model to pre-screen a massive number of candidate points in milliseconds, a large number of low-potential areas are filtered out. This makes mathematically rigorous systematic exploration, which was originally infeasible due to excessive computational costs, possible. The scheme evaluation time is shortened from hours to milliseconds, enabling the systematic exploration of a large-scale solution space under complex terrain to be completed within an engineering-acceptable timeframe. This solves the problem that the calculation of tower foundation costs and other factors under differentiated geological conditions based on digital terrain models is extremely time-consuming when performing high-precision engineering cost simulations of existing transmission lines. When this high-fidelity cost model is embedded into traditional optimization algorithms (such as genetic algorithms and simulated annealing algorithms) for iterative search, the search can only be performed in a very small local area of ​​the solution space, which is prone to getting trapped in local optima. It is impossible to fully explore the global solution space within the transmission line planning area, and it is impossible to guarantee that the quality of the final scheme is close to the theoretical global optimum. The excessively long computation time also makes it difficult to meet the timeliness requirements of engineering design. The line performance score prediction model makes global optimization in complex scenarios not only theoretically feasible but also practically usable, improving the generation efficiency of transmission line selection.

[0050] Step 3.4: Verify the candidate planning parameters that have passed the pre-screening based on the total cost of the route path. The candidate planning parameters that have been successfully verified are moved from the outer approximation set to the inner approximation set. Based on the latest outer approximation set and the latest inner approximation set, repeat the pre-screening and verification until convergence, and determine the converged inner approximation set.

[0051] Furthermore, step 3.4 includes: Step 3.4.1: Based on the candidate planning parameters selected through pre-screening, the starting point and ending point of the transmission line constitute the candidate path, calculate the total path cost of the candidate path.

[0052] Step 3.4.2: If the total cost of the candidate path is not greater than the set cost threshold, the candidate planning parameters are deemed to have been successfully verified, and the candidate planning parameters are moved from the outer approximation set to the inner approximation set.

[0053] Step 3.4.3: If the total cost of the candidate path is greater than the set cost threshold, the candidate planning parameter verification is determined to be unsuccessful. The external approximation set is updated based on the total cost of the candidate path and the planning index of the unverified candidate planning parameters.

[0054] The out-of-line approximation set is updated based on the total cost of the route path and planning indicators of the candidate planning parameters that failed validation, as expressed by the following formula: (7) In the formula, This represents the updated outer approximation set, specifically used in the (r+1)th iteration. Based on the total cost of the failed candidate planning parameters and the minimum distance from those parameters to the inner approximation set, the cutting surface of the outer approximation set is dynamically adjusted. The updated outer approximation set utilizes the known high cost of the failed points, local terrain information, and the exploration value of those points relative to the known high-quality solution set (inner approximation set) to dynamically construct a cost prediction cutting surface. This surface preemptively excludes areas where the predicted cost might exceed the limit (i.e., invalid areas) from the future search range, thereby guiding the algorithm to focus limited computational resources on more promising transmission lines. express Planning parameters in Let the outer approximate set be updated in the r-th iteration. This represents the total cost of the route path corresponding to the candidate planning parameters that failed validation. Indicates the exploration coefficient. Planning parameters indicating failed validation Planning indicators Representing the inner approximate set Planning parameters in Indicates the total cost of the route path. The gradient vector at that point, This represents the transpose operator. This represents the theoretically optimal cost. Indicates cost margin.

[0055] Step 3.4.4: Determine the latest candidate planning parameters and planning indices at the latest candidate planning parameters based on the latest outer approximation set and the latest inner approximation set. If the planning indices at the candidate planning parameters are greater than the set convergence threshold, it is determined that the convergence has not occurred. Then, repeat the pre-screening and verification of the latest candidate planning parameters and start the next iteration until the converged inner approximation set is obtained.

[0056] In Example 1, the iterative convergence mechanism of the ORACLE framework (i.e., step 3.1, active exploration sampling based on maximizing the minimum distance, and step 3.3, update of the outer approximate set cutting plane based on high-precision verification feedback) mathematically guarantees the asymptotic completeness of the exploration, ensuring that the final solution set can fully cover and compactly approximate the true global near-optimal space. This fundamentally guarantees the global near-optimal nature of the solution set, solving the problem in existing intelligent transmission line route selection where the blind and unsystematic search of the optimization algorithm leads to local traps in complex terrain (such as finding only a single obvious path along a ridge or valley), failing to guarantee finding a globally near-optimal route corridor. A comprehensive assessment of the entire feasible space may overlook technical issues related to alternative route selection solutions that are superior in terms of cost and construction difficulty. This invention, through the iterative convergence mechanism of the ORACLE framework, ensures that each output line undergoes rigorous high-precision cost verification. This enhances the global optimization capability of intelligent transmission line route selection in complex environments and the reliability of the route selection solution set. It provides multiple feasible implementation schemes within a preset cost margin but with different spatial orientations, improving the flexibility of decision support. It also reduces the risk of potential losses in line engineering costs due to insufficient optimization search, the risk of later design changes due to a single and unproven solution, and the significant external coordination risks that may arise from rigid route selection.

[0057] Step 4: Cluster the converged inner approximate set, select the planning parameter closest to the cluster center from each cluster to generate the final planning parameter set, and generate the final three-dimensional overhead transmission line through the path generator.

[0058] Embodiment 2 of the present invention provides an intelligent line selection system for overhead transmission lines, which implements the intelligent line selection method for overhead transmission lines described in Embodiment 1, including: The inner and outer approximation set generation module is used to obtain the terrain environment data of the transmission line planning area and the planning parameters of the transmission line. In the feasible space of the planning parameters, the optimal feasible solution is solved with the goal of minimizing the total cost of the line path. The optimal feasible solution is used as the inner approximation set, and the feasible space is used as the outer approximation set. The model building module is used to build a line performance score prediction model based on deep neural networks. The solution module uses the minimum distance between the outer approximation set and the inner approximation set as the planning index. It searches for candidate planning parameters in the outer approximation set that maximize the planning index, inputs the candidate planning parameters into the line performance score prediction model to obtain the predicted performance score, pre-screens candidate planning parameters based on the predicted performance score, and verifies the pre-screened candidate planning parameters based on the total cost of the line path. The verified candidate planning parameters are moved from the outer approximation set to the inner approximation set. Based on the latest outer approximation set and the latest inner approximation set, the pre-screening and verification are repeated until convergence, and the converged inner approximation set is determined. The route generation module is used to cluster the converged inner approximate set, select the planning parameters closest to the cluster center from each cluster to generate the final planning parameter set, and generate the final three-dimensional overhead transmission line through the path generator.

[0059] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0060] Finally, 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 the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A smart line selection method for overhead transmission lines, characterized in that: Obtain topographic environmental data and transmission line planning parameters of the planned area. Within the feasible space of the planning parameters, with the goal of minimizing the total cost of the line path, solve for the optimal feasible solution. Use the optimal feasible solution as the inner approximation set and the feasible space as the outer approximation set. A line performance score prediction model is constructed based on deep neural networks; Using the minimum distance between the outer approximation set and the inner approximation set as the planning index, candidate planning parameters are searched in the outer approximation set to maximize the planning index. The candidate planning parameters are input into the line performance score prediction model to obtain the predicted performance score. Candidate planning parameters are pre-screened based on the predicted performance score. The pre-screened candidate planning parameters are verified based on the total cost of the line path. The verified candidate planning parameters are moved from the outer approximation set to the inner approximation set. Based on the latest outer approximation set and the latest inner approximation set, the pre-screening and verification are repeated until convergence, and the converged inner approximation set is determined. Cluster the converged inner approximate set, select the planning parameter closest to the cluster center from each cluster to generate the final planning parameter set, and generate the final three-dimensional overhead transmission line through the path generator.

2. The intelligent line selection method for overhead transmission lines according to claim 1, characterized in that: The feasible space for generating planning parameters includes: Transmission lines are represented by B-spline curves containing multiple control points. The starting and ending points of the transmission lines are fixed as the first and last control points of the B-spline curve. The horizontal and vertical offsets of the remaining control points relative to the line connecting the first and last control points are set as planning parameters. Obtain topographic environmental data of the transmission line planning area, use the topographic environmental data to construct minimum clearance constraints, maximum slope constraints and obstacle avoidance distance constraints, determine the set of all planning parameters that satisfy the constraints, and generate the feasible space of the planning parameters.

3. The intelligent line selection method for overhead transmission lines according to claim 2, characterized in that: Using the optimal feasible solution as the inner approximation set and the feasible space as the outer approximation set, the following are included: The path generator uses planning parameters in the feasible space and the starting and ending points of the transmission line to generate a three-dimensional path, and repeatedly samples the three-dimensional path to obtain multiple coordinate sequences. Calculate the total cost of the route corresponding to each coordinate sequence; The planning parameters corresponding to the minimum total cost of the transmission route are taken as the optimal feasible solution. The optimal feasible solution and the starting and ending points of the transmission line constitute an inner approximate set. The planning parameters other than the optimal feasible solution in the feasible space and the starting and ending points of the transmission line constitute an outer approximate set.

4. The intelligent line selection method for overhead transmission lines according to claim 1, characterized in that: Acquire historical data of transmission lines, including terrain and environmental data and planning parameters for historical transmission line tasks; The historical data of transmission lines and their corresponding total route costs are standardized and used as the sample set. The standardized total route costs are then negative and set as the true labels. Based on the sample set and the real labels, a deep neural network is trained by minimizing the mean square error between the predicted line performance score and the real label to obtain the line performance score prediction model.

5. The intelligent line selection method for overhead transmission lines according to claim 1, characterized in that: The minimum distance between the outer approximation set and the inner approximation set is used as the planning index. Candidate planning parameters are searched in the outer approximation set to maximize the planning index, as expressed by the following formula: In the formula, Indicates the candidate planning parameters. express Maximize the planning index at the location, Describing the outer approximation set Planning parameters to inner approximate set Planning parameters The minimum distance is the planning parameter. Planning indicators Represents Euclidean distance. This means that all constraints are satisfied. of The goal is to find the planning parameters that maximize the planning indicators as candidate planning parameters.

6. The intelligent line selection method for overhead transmission lines according to claim 5, characterized in that: Candidate planning parameters are pre-screened based on predicted performance scores, including: The candidate planning parameters are input into the line performance score prediction model, the predicted performance score is output, and the performance threshold is determined based on the maximization planning index of the candidate planning parameters. The predicted performance score is compared with a performance threshold to determine the candidate planning parameters that pass the pre-screening, including: If the predicted performance score is not less than the performance threshold, the candidate planning parameters are deemed to have passed the pre-screening. If the predicted performance score is less than the performance threshold, the candidate planning parameters are determined to have failed the pre-screening. Based on the set relaxation factor and the candidate planning parameters that have passed the pre-screening, new candidate planning parameters are constructed to replace the candidate planning parameters that have failed the pre-screening, thereby updating the outer approximation set. In the latest outer approximation set, the candidate planning parameters are re-determined to maximize the planning index.

7. The intelligent line selection method for overhead transmission lines according to claim 6, characterized in that: The performance threshold is determined based on the planning indicators of the candidate planning parameters, expressed by the following formula: In the formula, Indicates the performance threshold. Indicates the baseline performance threshold. Indicates the exploration coefficient. It represents the maximization of planning index based on the candidate planning parameters, which are the inner approximation set with the optimal feasible solution and the outer approximation set with the feasible space.

8. The intelligent line selection method for overhead transmission lines according to claim 1, characterized in that: The convergent inner approximation set includes: Based on the candidate planning parameters selected through pre-screening, and the starting and ending points of the transmission lines to form candidate paths, the total route cost of the candidate paths is calculated. If the total cost of the candidate path is not greater than the set cost threshold, the candidate planning parameter is deemed to have been successfully verified, and the candidate planning parameter is moved from the outer approximation set to the inner approximation set. If the total cost of a candidate path exceeds the set cost threshold, the candidate planning parameter verification is deemed to have failed. The external approximation set is then updated based on the total cost of the candidate path and the planning indicators of the failed candidate planning parameters. The latest candidate planning parameters and planning indices at the latest candidate planning parameters are determined based on the latest outer approximation set and the latest inner approximation set. If the planning indices at the candidate planning parameters are greater than the set convergence threshold, it is determined that the convergence has not been achieved. The latest candidate planning parameters are then pre-screened and verified repeatedly, and the next iteration begins until the converged inner approximation set is obtained.

9. The intelligent line selection method for overhead transmission lines according to claim 7, characterized in that: The out-of-line approximation set is updated based on the total cost of the route path and planning indicators of the candidate planning parameters that failed validation, as expressed by the following formula: In the formula, Represents the updated outer approximation set. express Planning parameters in Let the outer approximate set be updated in the r-th iteration. This represents the total cost of the route path corresponding to the candidate planning parameters that failed validation. Indicates the exploration coefficient. Planning parameters indicating failed validation Planning indicators Representing the inner approximate set Planning parameters in Indicates the total cost of the route path. The gradient vector at that point, This represents the transpose operator. This represents the theoretically optimal cost. Indicates cost margin, It represents the maximization of planning index based on the candidate planning parameters, which are the inner approximation set with the optimal feasible solution and the outer approximation set with the feasible space.

10. An intelligent route selection system for overhead transmission lines, operating the intelligent route selection method for overhead transmission lines according to any one of claims 1-9, characterized in that: The inner and outer approximation set generation module is used to obtain the terrain environment data of the transmission line planning area and the planning parameters of the transmission line. In the feasible space of the planning parameters, the optimal feasible solution is solved with the goal of minimizing the total cost of the line path. The optimal feasible solution is used as the inner approximation set, and the feasible space is used as the outer approximation set. The model building module is used to build a line performance score prediction model based on deep neural networks. The solution module uses the minimum distance between the outer approximation set and the inner approximation set as the planning index. It searches for candidate planning parameters in the outer approximation set that maximize the planning index, inputs the candidate planning parameters into the line performance score prediction model to obtain the predicted performance score, pre-screens candidate planning parameters based on the predicted performance score, and verifies the pre-screened candidate planning parameters based on the total cost of the line path. The verified candidate planning parameters are moved from the outer approximation set to the inner approximation set. Based on the latest outer approximation set and the latest inner approximation set, the pre-screening and verification are repeated until convergence, and the converged inner approximation set is determined. The route generation module is used to cluster the converged inner approximate set, select the planning parameters closest to the cluster center from each cluster to generate the final planning parameter set, and generate the final three-dimensional overhead transmission line through the path generator.

Images