Local adjustment method and system for arrangement of power transmission line towers

Abstract

The invention discloses a local adjustment method and system for electric transmission line tower arrangement, belongs to the technical field of electric transmission line design, and aims to solve the problems of local violation, high global optimization cost, lack of a self-adaptive repair strategy and dependence on experience decision in traditional manual arrangement. The method comprises the following steps: firstly, carrying out conductor sag fitting and comprehensive checking of sag, ground clearance, crossing and windage yaw on a tower arrangement scheme; recognizing a violation tower position according to a result, and determining a local repair section; then generating candidate repairing actions such as tower moving, tower height increasing and tower inserting; then constructing a multi-objective optimization model, quantifying the repair cost and sequencing the actions; and finally, executing the action according to the priority, and iteratively checking until the hard constraint is met. The method focuses on local repair, reduces the cost, improves the efficiency, adapts to multiple voltage grades and complex terrains, and provides intelligent support for pole tower arrangement design.

Description

Technical Field

[0001] This invention relates to the field of power transmission line design technology, and specifically to a method for local adjustment of the arrangement of power transmission line towers. Background Technology

[0002] The design of transmission line tower placement requires comprehensive optimization of tower location, tower type selection, and span configuration, while simultaneously meeting multiple mandatory safety constraints such as ground clearance, wind clearance, safety when crossing obstacles, and line breakage conditions. As transmission line projects expand into complex environments such as mountainous areas, canyons, and urban intersections, traditional manual placement and static verification methods have revealed numerous problems. Ranking schemes are prone to local non-compliance with constraints. Especially in complex terrain areas such as sags, clear spaces, and crossings, violations are discovered late and at a high cost.

[0003] A complete overhaul would be extremely costly: any violation could lead to a complete reshuffling of the rankings, which would be time-consuming and costly.

[0004] Lack of adaptive repair strategies oriented towards engineering costs: The factors affecting ranking are diverse, and there is a lack of an accurate feedback mechanism for the actual cost.

[0005] Repair actions rely on experience and lack intelligent decision-making: operations such as moving towers, raising towers, and installing towers lack systematic prioritization and optimization evaluation.

[0006] Therefore, there is an urgent need for an intelligent ranking technology that combines precise verification, local intelligent repair, and global coupled control to achieve simultaneous optimization of safety and economy. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for local adjustment of the arrangement of transmission line towers. This method can accurately identify local constraint violations and intelligently generate the optimal repair scheme, thereby reducing repair costs and improving repair efficiency while meeting safety constraints.

[0008] To achieve the above objectives, the technical solution of the present invention is as follows: A method for local adjustment of the arrangement of transmission line towers, characterized by comprising the following steps: 1) Fit conductor sag to the generated tower placement scheme to obtain the conductor position distribution along the line. Then, make a comprehensive judgment on sag check, ground clearance check, crossing check and wind deflection check. The comprehensive judgment needs to first confirm whether the single check dimension meets its respective safety standard, and then confirm whether the multi-dimensional constraints meet the standards at the same time. If any dimension does not meet the standard, it is judged as a violation and triggers the subsequent local repair process. If all dimensions meet the standard, it is judged as qualified and no repair is required. 2) Based on the verification results of step 1), identify the tower locations that do not meet the hard constraints, and determine the local repair section consisting of the tower location and its adjacent tower locations before and after it. The hard constraints include sag verification, ground clearance verification, cross-span verification, and wind deflection verification. The length of the local repair section is dynamically set within the range of 200m to 2000m according to the distribution of the violation constraints. 3) For the local repair segment determined in step 2), automatically generate multiple candidate repair actions, including tower relocation, tower height increase, tower insertion, and local segment rearrangement. Each candidate repair action corresponds to at least one adjustment parameter, which includes: Adjustment parameters for tower relocation: horizontal offset of the target tower location along or perpendicular to the line direction, with final coordinates such as latitude and longitude / cumulative distance; the offset must be within ±20m to 50m of the center of the target tower location, and must match the discretized search results of DEM elevation and slope data.

[0009] Tower height adjustment parameters: tower height increment or final height, tower type level such as the upgraded tower type model; structural limitations such as rotation level and maximum horizontal span need to be checked simultaneously to ensure that the sag and clearance constraints are met after adjustment.

[0010] Tower placement adjustment parameters: tower placement location coordinates such as cumulative distance / latitude and longitude, adjacent span allocation ratio or specific span value; parameters must meet span compliance requirements and are determined after calculating the cumulative true marginal cost through dynamic programming.

[0011] The adjustment parameters for the partial segment rearrangement action are: the relative position offset of 3 to 5 towers, and the span combination after reorganization, including the span values ​​between each adjacent tower; it is necessary to ensure that the span difference after reorganization does not exceed the design specifications and meets the construction feasibility constraints.

[0012] 4) Construct a multi-objective local optimization model to meet hard constraints. Quantify the repair costs of different candidate repair actions based on the true marginal cost. Then, combine the optimization strategy to rank the candidate repair actions and output the ranking results. The true marginal cost is determined by the weighted superposition of the cost increment of the project along the line and the cost increment of the tower node. 5) Perform the repair action according to the sorting priority output in step 4), and re-execute steps 1) to 4) on the adjusted local solutions until all local solutions satisfy the hard constraints.

[0013] Furthermore, in step 1), the sag check needs to determine whether the geometric shape of the conductor is reasonable under different meteorological conditions, whether the conductor tension and stress level are within a safe and controllable range, whether the sag size meets the design threshold, and whether the sag data is suitable for the calculation requirements of subsequent clearance check and wind deflection check. The check is based on the conductor mechanical safety code, design tension and sag threshold and structural adaptability requirements. Furthermore, the ground clearance check first calculates the sag based on a parabola approximation. When the minimum clearance is less than the configured threshold, it switches to the catenary explicit solver to obtain the precise coordinates of the lowest point. Then, it obtains the sag curve of the conductor through the catenary or piecewise fitting model, searches for the lowest point of sag within the continuous span range, extracts the ground DEM or ground feature model, and calculates the difference between the height of the conductor at the cumulative distance x position on the sag curve and the elevation of the highest point of the ground or ground feature at the corresponding position as the clearance value. If the clearance value is less than the set standard, the check is not satisfied. The crossing verification needs to determine whether the vertical clearance and horizontal offset distance between the lowest point of the conductor and the highest point of the crossing object meet the standards. In high-risk scenarios, it is additionally determined whether the electric field influence is compliant. For sensitive crossing objects, the verification threshold is raised according to the "penalty coefficient". The verification basis is the crossing safety distance requirements in the national standard and the penalty coefficient correction threshold corresponding to the risk level. The wind deflection check is based on the maximum allowable wind deflection angle of the selected tower insulator string as the control index. It calculates the lateral offset curve of the conductor under wind load by combining the reference wind speed, air density, terrain roughness and wind load amplification factor of the meteorological zone where the line is located. It judges whether the safety clearance between the conductor and ground objects, crossing targets and adjacent phase conductors after the offset meets the standard, and whether the swing angle of the insulator string exceeds the maximum allowable swing angle. The verification basis is the maximum allowable wind deflection angle of the tower insulator string, local meteorological parameters and terrain correction factor.

[0014] Furthermore, in step 3), the tower moving action is carried out by discretizing the DEM elevation and slope data within a range of ±20m to 50m from the center of the target tower location, and combining the construction feasibility constraints and hard constraints such as sag, clearance, wind deflection, and cross-span to calculate the real marginal cost and horizontal disturbance score of each candidate tower location, and selecting the candidate tower location with the highest comprehensive score as the final tower moving location. The tower insertion action adopts a dynamic programming method. Within the local repair section, the cumulative real marginal cost of different candidate tower positions is calculated under the constraints of span, sag, clearance, wind deflection and cross-span hard constraints. The optimal tower insertion position and adjacent span combination are generated by backtracking. The local segment rearrangement action combines and transforms local tower segments containing 3 to 5 towers, changes the relative positions and span combinations of adjacent towers, performs sag, clearance, wind deflection, cross span and construction feasibility checks, and selects the optimal tower combination from the candidate combinations that satisfies hard constraints and has the lowest actual marginal cost. When the tower height increase action automatically selects a higher-level tower type or a special corner tower type within the set range, it verifies the corner level, maximum horizontal span, and call height conditions in real time. Combining the results of local sag, headroom, wind deflection, and cross-span verification, it selects the optimal tower type that meets the hard constraints and has the lowest actual marginal cost.

[0015] Furthermore, in step 3), when generating candidate repair actions, a repair action generation strategy tree is constructed based on the verification feedback results. This repair action generation strategy tree includes action nodes, decision nodes, and terminal nodes. The action nodes are tower relocation, tower insertion, and tower height increase. The decision nodes are constraint satisfaction. The branch design of the repair action generation strategy tree predefines action priorities, and each node is bound to three types of node indicators: real marginal cost, construction feasibility, and local disturbance. After each local action is executed, the repair action generation strategy tree is dynamically updated based on the verification results. This strategy tree is combined with global ranking or other local repair actions to form a local-global closed-loop optimization system. At the same time, the nodes save the original scheme and simulation results, supporting manual review or automatic rollback.

[0016] Furthermore, in step 3), the repair action and verification feedback are linked to generate a strategy tree. The method for generating the strategy tree includes: Based on the comprehensive verification results of step 1), the violation type, constraint compliance status and local repair segment parameters are extracted to determine the initial input conditions for the construction of the strategy tree; Define the node structure of the strategy tree, where action nodes correspond to three core repair actions: moving towers, inserting towers, and increasing tower height; decision nodes are hard constraint satisfaction judgments; and terminal nodes are feasible repair solution outputs. Establish the mapping relationship between nodes and repair actions. The design strategy tree branch structure is used to preset the action priority of safety over economy in the branches, and bind three types of indicators for each action node: real marginal cost, construction feasibility and local disturbance, and clarify the quantitative rules of the indicators. Trigger a branch traversal of the policy tree, match the initial input conditions with the action nodes, generate corresponding candidate repair actions, and simulate the execution of the action; The decision node determines whether the simulation execution meets the hard constraints of sag, ground clearance, crossing, and wind deflection. If it does, the action and parameters are output to the terminal node to form a feasible solution. If it does not meet the constraints, the indicator weights of the action node are adjusted based on the new verification feedback results, and candidate repair actions are rematched. After each local repair action is completed, the node data, branch priority, and action parameters of the strategy tree are updated synchronously, so that the strategy tree is linked with the global ranking or other local repair actions to form a local-global closed-loop optimization system. Each node saves the original scheme and simulation results, supporting manual review or automatic rollback.

[0017] Furthermore, in step 4), the Pareto front ordination is used to sort the candidate repair actions to filter non-dominated solutions, specifically including the following steps: a) Define the optimization objectives and indicator quantification rules. The optimization objectives include economy, engineering stability and construction feasibility. Economy is quantified by real marginal cost, engineering stability is quantified by local disturbance, and construction feasibility is quantified by terrain adaptability and construction period adaptability. It is also clear that safety takes precedence over economy. Candidate repair actions that do not meet hard constraints are directly excluded from the Pareto front. b) Generate all possible candidate repair actions for the local repair segment, form a candidate action pool, and calculate the economic efficiency, engineering stability and construction feasibility index values ​​for each candidate repair action; c) Iteratively filter the candidate action pool according to the Pareto dominance relation, eliminate dominated solutions, and the remaining non-dominated solutions constitute the Pareto front. The Pareto dominance relation is: if candidate repair action A is not inferior to candidate repair action B on all optimization objectives, and is strictly superior to B on at least one optimization objective, then A dominates B. d) Determine the priority of non-dominated solutions on the Pareto front according to engineering requirements, and sort the non-dominated solutions by weight based on a dynamic weight adjustment mechanism, which includes three methods: predefined strategy table, real-time dynamic update and adaptive optimization.

[0018] Furthermore, in step d), the predefined strategy table establishes a weight table based on the project stage, line level, and site type, and the corresponding weight is automatically called before each local repair action is calculated; the real-time dynamic update adjusts the weight in real time according to the verification results, construction restrictions, or cost changes, and links with the local repair strategy tree to affect the selection order of action nodes; the adaptive optimization automatically fine-tunes the weights according to the repair effect through a feedback mechanism to ensure the optimality and stability of the selection of local repair actions.

[0019] Furthermore, in step 4), the output sorting results include repair action instructions, changes to tower location range, cost predictions, and verification result descriptions. The repair action instruction includes the action type, target tower number, position adjustment amount, tower type adjustment information, and action execution order; The scope of the changed tower location includes the affected tower location segment, the original tower location coordinates and the adjusted coordinates, the adjustment range and a description of the disturbance. The cost forecast includes the actual marginal cost of a local segment, the segment cost increment, and the node cost increment. The verification results description includes verification indicators and verification status markers for each repair action, and provides detailed data tables or visualization reports.

[0020] Furthermore, in scenarios where the line crosses railways, buildings, rivers, and other important features, a dedicated penalty coefficient is introduced to dynamically raise the clearance, wind deflection, and safety clearance verification thresholds. The penalty coefficient is iteratively optimized by accumulating data related to the penalty coefficients and safety accident rates of similar projects, and the corrected standard limit is calculated as: basic limit × penalty coefficient.

[0021] Furthermore, when multiple local segments fail to meet the constraints, parallel computing is enabled to accelerate the repair efficiency, specifically including the following steps: i) Calculate the influence domain for each segment to be repaired. The influence domain covers the segment to be repaired and the adjacent towers that may be affected by the repair action. The calculation of the influence domain first determines the initial boundary based on the local repair segment, then expands to the adjacent towers according to the hard constraint association rule, and finally dynamically adjusts the range in combination with the repair action type. ii) Construct a conflict graph based on the overlapping relationship of the influence domains of each segment to be repaired, and divide the segments to be repaired into mutually exclusive parallel batches by graph coloring or the maximum independent set algorithm to ensure that the operations of the segments to be repaired within the same batch do not conflict with each other. iii) Within each parallel batch, the repair action trial is performed on the local data snapshot for each segment to be repaired, including candidate action generation, sag / clearance / wind deflection / crossing verification and real marginal cost assessment, and candidate repair actions are evaluated in parallel within the segment; iv) After completing the trial calculation, perform atomic updates and detect conflicts through an optimistic concurrent commit mechanism. If a conflict in the affected domain is detected, perform a rollback or rescheduling to ensure global data consistency.

[0022] Furthermore, in step iv), the optimistic concurrent commit mechanism includes the following steps: Each segment to be repaired executes a complete repair process on a dedicated local data snapshot, generating a local solution after repair. The trial calculation process does not modify the original data of the global tower placement model. After all parallel segments have completed the trial calculations, the overlap of the influence domains and data consistency of the local solutions after each segment are compared to determine whether there are any conflicts. If there are no conflicts, the repaired local solutions for each segment will be synchronously updated to the global tower placement model, and the repair log will be recorded. If a conflict exists, the submission process will be terminated, the conflicting segments will be rolled back to the original snapshot state before the trial calculation, the parallel batches will be re-divided or the repair action parameters will be adjusted, and the trial calculation and conflict detection will be restarted until there are no conflicts before submission.

[0023] Furthermore, when multiple local segments do not satisfy the hard constraints, the repair efficiency is accelerated by combining parallel computing with an optimistic concurrent commit mechanism. This accelerated repair includes: For each segment to be repaired, the initial boundary is first determined based on the segment, then extended to the adjacent towers according to the hard constraint association rules, and finally the range is dynamically adjusted in combination with the repair action type to obtain the influence domain covering the segment to be repaired and the adjacent towers that may be affected. A conflict graph is constructed based on the overlapping relationship of the influence domains of each segment to be repaired. By using graph coloring or the maximum independent set algorithm, the non-conflicting segments to be repaired are divided into mutually exclusive parallel batches to ensure that the operations between segments within the same batch do not interfere with each other. Within each parallel batch, the repair segment is independently calculated on a dedicated local data snapshot, including candidate repair action generation, sag / clearance / wind deflection / crossing verification and real marginal cost assessment, and candidate repair actions are evaluated in parallel within the segment; After all parallel batches have completed the trial calculations, the overlap of the influence domains and data consistency of the local solutions after each repair are compared to determine whether there are any conflicts. If there are no conflicts, the local solutions after each repair will be synchronously updated to the global tower ranking model, and the repair log will be recorded. If there are conflicts, the submission process will be terminated, the conflicting segments will be rolled back to the original snapshot state before the trial calculation, the parallel batches will be re-divided or the repair action parameters will be adjusted, and the trial calculation and conflict detection will be restarted until there are no conflicts before submission.

[0024] Furthermore, before performing the repair action, a data snapshot is generated for the affected local segment and its adjacent affected area. The data snapshot includes tower location coordinates, tower type parameters, span information, sag curve, verification index and local cost surface. Simulate, verify, and calculate costs for candidate actions on a local snapshot. Select the optimal repair action to try and apply it. If the verification or constraint validation fails, restore the local segment to its original state using the data snapshot, clear the impact of the attempted repair actions on the global model, and mark the candidate action as infeasible or adjust its parameters for the next round of repair action generation strategy.

[0025] Furthermore, during the remediation process, a data snapshot rollback mechanism ensures the consistency of the global solution, and the steps are executed sequentially according to the following logic: Before initiating the local repair process, a complete data snapshot containing tower location coordinates, tower type parameters, span information, sag curves, verification indicators, and local cost surfaces is generated for the affected local section and its adjacent affected area to preserve the original scheme benchmark. A local simulation environment is created based on the data snapshot. In this environment, candidate repair action simulation, sag / clearance / wind deflection / crossing verification and real marginal cost calculation are performed independently without modifying the original data of the global tower ranking model. After the simulation is completed, it is determined whether the repair action meets the hard constraints and meets the requirements of economy and construction feasibility: if it does, the repair action is applied to the global model; if it does not, the rollback process is triggered. When the rollback process is executed, the generated data snapshot is called to restore all parameters of the affected local segment and the adjacent affected domain to their original state before the repair action was executed, thus eliminating the potential impact of the invalid repair action on the global model. After the rollback is completed, the candidate repair action is marked as infeasible, or its parameters are adjusted based on the verification feedback and then reintroduced into the next round of repair action generation strategy. At the same time, the rollback log and candidate action evaluation results are recorded for subsequent strategy optimization and manual review.

[0026] This invention also discloses a system for local adjustment of the tower placement of transmission lines. The system is applicable to voltage levels from 35kV to 1000kV and different meteorological zones, and is automatically coupled with a global tower placement optimization module to form a full-line intelligent closed-loop placement system. Specifically, it includes the following steps: The global ranking generation module generates an initial global ranking scheme based on the line start and end points, tower type library, and engineering constraints. The global optimization module optimizes the initial global ranking scheme based on the overall cost surface, engineering constraints, and construction feasibility to obtain global candidate ranking schemes. The local repair module receives global candidate ranking schemes and performs repair action generation, Pareto filtering, real marginal cost evaluation, and optimal action application for local segments that do not meet the constraints. After a local repair is completed, the repair results, cost surface updates, and verification feedback information are sent back to the global optimization module as input for the next round of global optimization or local repair strategy adjustment, until all towers along the entire line meet safety, economic, and construction constraints, generating the final overall ranking scheme, and retaining the local repair history, iteration logs, and cost assessment data.

[0027] The beneficial effects of this invention are as follows: This invention achieves localized intelligent adjustment of transmission line tower placement through a complete process of "verification and identification - repair section delineation - action selection generation - optimization sorting - iterative repair - closed-loop control". It focuses on targeted repairs of local constraint violation areas, avoiding the high cost of global re-optimization; through real marginal cost quantification and optimal strategy sorting, it ensures that the repair scheme has optimal economic efficiency while meeting safety constraints; and by combining data snapshot rollback, parallel computing, and a global closed-loop coupling mechanism, it improves the controllability, efficiency, and overall optimality of the repair process. This method is applicable to different voltage levels and complex terrain environments, effectively solving many drawbacks of traditional manual placement and static verification, and providing efficient and intelligent technical support for transmission line tower placement design. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the local repair strategy tree of the present invention.

[0029] Figure 2 This is the local Pareto set filtering process of the present invention.

[0030] Figure 3 This is a geometrical diagram of the sag and clearance of the present invention.

[0031] Figure 4 This is a schematic diagram illustrating the tower displacement process implemented by the optimal point search procedure near the DEM in this invention.

[0032] Figure 5 This is a schematic diagram of the local segment parallel repair process of the present invention.

[0033] Figure 6 This is a global-local closed-loop framework diagram of the present invention. Detailed Implementation

[0034] To make the purpose, technical solution, and advantages of the invention clearer, the invention will be further described below with reference to the accompanying drawings.

[0035] As shown in the figure, the present invention provides a method for local adjustment of the arrangement of transmission line towers, comprising: Step 1: Fit conductor sag to the generated tower placement scheme, obtain the conductor position distribution along the line, and make a comprehensive judgment on sag check, ground clearance check, crossing check and wind deflection check; In this invention, the comprehensive judgment refers to determining whether the generated tower placement scheme meets all mandatory safety constraints. Specifically, it focuses on the compliance of each verification dimension, ultimately clarifying whether there are violations in a local segment and whether repair needs to be initiated. The specific judgment content is as follows: The core of the comprehensive judgment is double confirmation, including: 1. Whether a single dimension meets its respective security standards, i.e., the exclusive constraint requirements for each type of verification; 2. Whether the constraints of multiple dimensions meet the standards simultaneously, with no violations in any dimension, to avoid the situation where a single dimension is compliant but the whole is unsafe. Finally, a clear conclusion of qualified or non-compliant is output. If qualified, no repair is required; if non-compliant, the subsequent local repair process is triggered.

[0036] The specific judgment criteria for sag verification, ground clearance verification, intersection verification, and wind deflection verification are shown in the table below:

[0037] All four verification dimensions must be met simultaneously for a comprehensive assessment to be considered satisfactory: First, basic data is obtained through sag verification to provide calculation input for the other three types of verification; Then, assess the compliance of ground clearance, crossing, and wind deflection separately; If any dimension fails to meet the requirements, such as insufficient clearance, excessive wind deflection, or insufficient crossing distance, it will be judged as a violation, and the violation point needs to be marked and a local repair section needs to be determined. If all dimensions are met, the overall judgment is that it is qualified, and the ranking scheme for this segment does not need to be repaired.

[0038] The essence of this invention is a multi-dimensional safety compliance verification. The core is to confirm whether the tower placement scheme meets the mandatory safety constraints of the country / industry through a full-coverage inspection of four dimensions, and finally determine whether local repairs need to be initiated, so as to avoid engineering risks caused by omissions in a single dimension.

[0039] (2) Identify the tower locations that do not meet the hard constraints based on the verification results, and determine the local repair section composed of the towers before and after them; (3) Automatically generate multiple candidate repair actions for local repair segments, including tower moving, tower height increase, tower insertion, and local segment rearrangement, with each action corresponding to at least one adjustment parameter; (4) Construct a multi-objective local optimization model, with hard constraints satisfied as a premise, quantify the repair cost of different candidate actions based on the real marginal cost, and sort the candidate actions by combining the optimization strategy and output the sorting results; (5) Perform the repair action according to the sorting priority, and re-execute steps (1) to (4) on the adjusted local scheme until all local schemes meet the hard constraints.

[0040] The hard constraints in step (2) include: sag check, ground clearance check, cross-span check, and wind deflection check.

[0041] The sag check is used to determine the geometric shape and mechanical safety of the conductor under different meteorological conditions, ensuring that both conductor tension and sag are within a controllable range. The stress method is used to calculate the tension transition between different conditions, considering the effects of conductor temperature changes, icing weight, and wind load. The check results must simultaneously satisfy the requirements for maximum sag, maximum tension and stress level, and structural adaptability. The sag check also serves as the basis for subsequent clearance and wind deflection checks.

[0042] The ground clearance check first calculates the sag based on a parabola approximation, and then switches to the catenary explicit solver to obtain the precise coordinates of the lowest point when the minimum clearance is less than the configured threshold.

[0043] Ground clearance verification method: Obtain the conductor sag curve based on the catenary or piecewise fitting model; Search for the lowest point of sag within the current continuous range; Extracting the ground DEM / ground feature model; Ground clearance:

[0044] in The cumulative distance on the sag curve is The height of the conductor at the location in The cumulative distance is Elevation of the highest point of the ground or feature at the location like:

[0045] The verification is not satisfied.

[0046] The specific steps are as follows: Discrete sampling points are set across the segment, the terrain elevation is read, the elevation value of the sag curve with respect to the cumulative distance is calculated, the clearance is calculated for each sampling point, and the minimum clearance value is used for ground clearance verification. If the condition is not met, it is marked as a violation point and included in the local repair segment; if the condition is met, the verification is passed.

[0047] The crossing verification is based on the clearance requirements between the lowest point of the conductor and the highest point of the crossing target under national standard operating conditions. It primarily verifies the safety distance requirements between the line and objects it crosses, such as railways, highways, buildings, water bodies, and communication lines. Vertical clearance, horizontal offset distance, and, if necessary, electric field influence verification must be considered.

[0048] The wind deflection verification step includes: calculating the unit wind load based on the given wind speed, and obtaining the conductor offset by approximate analytical or numerical methods. In this invention, the wind deflection is simplified, and the maximum allowable angle of the tower string is used as the wind deflection angle limit.

[0049] The wind deflection check is based on the maximum allowable wind deflection angle of the selected tower insulator string as the control index, and calculates the lateral deflection curve of the conductor under wind load by combining the reference wind speed, air density, terrain roughness and wind load amplification factor of the meteorological zone where the line is located. The safety gap between the conductor and ground objects, crossing targets and adjacent phase conductors after the deflection is checked to ensure that the electrical and structural safety requirements under the maximum wind deflection condition meet the specifications.

[0050] The lateral offset calculation takes into account the following factors: wind load parameters of the meteorological zone where the project is located, actual tower string length and structural layout, unit load of conductor, self-weight and tension state, and local wind speed correction coefficient caused by terrain.

[0051] Criteria for wind deflection check: String swing angle limit verification

[0052] in The maximum allowable swing angle of the insulator string Offset leads to minimum safety gap check

[0053] The ground clearance check is based on a segmented comparison between the lowest point of the traverse sag curve and the elevation data of ground features along the route, calculating the elevation difference between the lowest point of the sag and the ground. Figure 3 The system checks whether the elevation difference meets the minimum safe clearance standard corresponding to different land cover types. If it does not meet the standard, the check fails and a local repair action is triggered.

[0054] In step (2), the length of the local repair segment is dynamically set within the range of 200m to 2000m based on the distribution of the violation of constraints.

[0055] In step (3), the tower insertion repair action uses a dynamic programming method to select the optimal tower insertion location and adjacent span combination within the repair section. The tower insertion action uses a dynamic programming method, which calculates the cumulative real marginal cost of different candidate tower locations within the local repair section using span, sag, clearance, wind deflection, and hard constraints of intersection, and generates the optimal tower insertion location and adjacent span combination through backtracking.

[0056] In step (3), the tower relocation action involves searching the DEM elevation and slope data within a range of ±20m to 50m from the center of the target tower location to determine the optimal feasible tower location. The tower relocation action involves discretizing the DEM elevation and slope data within a range of ±20m to 50m from the center of the target tower location, combining construction feasibility constraints with hard constraints on sag, clearance, wind deflection, and cross-span, calculating the true marginal cost and horizontal disturbance score of each candidate tower location, and selecting the candidate tower location with the highest comprehensive score as the final tower relocation location.

[0057] In step (3), the segment rearrangement action combines and transforms local tower location segments containing 3 to 5 towers and performs feasibility checks. The segment rearrangement action combines and transforms local tower location segments containing 3 to 5 towers by changing the relative positions and span combinations of adjacent towers, and performs checks on sag, clearance, wind deflection, cross-span, and construction feasibility. The optimal tower location combination that satisfies hard constraints and minimizes actual marginal cost is selected from the candidate combinations. Specifically, the segment length is first set based on the positions of 3 to 5 nearby towers, then adaptive discretization is performed according to different slopes, and multi-objective optimization iteration is conducted by combining hard constraint satisfaction, economic optimization, and disturbance minimization. Finally, the optimal combination is output.

[0058] In step (3), the tower type upgrade action, such as increasing tower height, automatically verifies structural limitations, including the corner level, maximum horizontal span, and height call conditions. When the tower type upgrade action automatically selects a higher-level tower type or a special corner tower type within the set range, it verifies structural limitations in real time, including the corner level, maximum horizontal span, and height call conditions. It also combines the results of local sag, clearance, wind deflection, and cross-span verification to select the optimal tower type that meets the hard constraints and has the lowest actual marginal cost.

[0059] In step (3), the local repair involves using the verification feedback results to construct a repair action generation strategy tree, which is used to dynamically deduce the optimal repair tower placement, such as... Figure 1 The strategy tree is used to dynamically derive the optimal tower placement for repair. For violations of hard constraints such as sag, clearance, wind deflection, and crossing, it systematically generates local repair schemes. The verification feedback-driven repair action strategy tree is designed to systematically and automatically generate local repair schemes after local verification (including sag, clearance, wind deflection, and crossing) reveals unmet hard constraints during transmission line tower placement optimization. Optimal tower placement for repair is achieved through dynamic derivation. First, action nodes, decision nodes, and terminal nodes are defined. Action nodes include tower relocation, tower insertion, and tower height increase; decision nodes include constraint satisfaction judgments; and terminal nodes include feasible solution outputs. Then, branch design is performed. In branch design, action priorities can be predefined (e.g., safety takes precedence over economy). Each node can be bound to three types of node indicators: formal marginal cost, construction feasibility, and local disturbance. After each local action is executed, the verification results can dynamically update the strategy tree; the decision tree, together with the global ranking or other local repair actions, forms a local-global closed-loop optimization system; the decision tree node can save the original scheme and simulation results, and supports manual review or automatic rollback.

[0060] The true marginal cost in step (4) is determined by a weighted sum of the incremental costs of the engineering along the route and the incremental costs of the tower location nodes, used to accurately reflect the impact of local adjustments on the project cost. The incremental costs of the engineering along the route include: the impact of changes in span distance, the construction complexity caused by changes in sag, and other costs. The incremental costs of the tower location nodes include: tower type and tower materials, incremental costs of foundation construction, and auxiliary and supporting costs.

[0061] In step (4), Pareto front sorting is used to screen non-dominated solutions for candidate actions. Figure 4In this invention, Pareto front ranking is a multi-objective optimization strategy used to select the optimal solution from multiple candidate repair actions such as tower relocation, tower insertion, and tower addition. Its core is to solve the trade-off between multiple objectives such as safety, economy, and construction feasibility, avoiding the optimization of a single objective from harming other objectives, such as pursuing only the lowest cost while ignoring safety constraints. In step (4), the candidate repair actions are Pareto front ranked at each node of the local policy tree to select the set of actions that are non-dominated solutions in terms of real marginal cost, local disturbance, and construction feasibility indicators, thereby providing input for dynamically deriving the optimal repair tower ranking. The Pareto front ranking process for selecting non-dominated solutions includes: candidate action generation, front selection rules, weighted ranking of Pareto front actions according to policy weights, and candidate action updating.

[0062] Pareto dominance: For two candidate repair actions A and B, if action A is not inferior to B in all optimization objectives, including lower real marginal cost, smaller local perturbation, and higher construction feasibility, and is strictly superior to B in at least one objective, then A dominates B, and B is a dominated solution and can be directly eliminated. If A and B each have advantages in different objectives, such as A having lower cost but larger perturbation, and B having higher cost but smaller perturbation, then they are mutually non-dominated solutions and are both retained to proceed to the next stage. In this invention, candidate actions on the Pareto front are either the lowest cost and satisfy safety constraints, or the smallest perturbation and controllable cost; there is no superoptimal solution that has lower cost, smaller perturbation, and higher safety. This invention uses Pareto front ranking for priority screening of candidate repair actions, which needs to combine three core indicators: real marginal cost, local perturbation, and construction feasibility. The specific process is strongly bound to the patent technology details and can be divided into four steps: 1. Define optimization goals and quantify indicators.

[0063] This invention prioritizes safety over economy. If a candidate action does not meet hard constraints such as insufficient headroom or excessive wind deflection, it is directly excluded from the Pareto front, regardless of how advantageous the cost or disturbance, to ensure that the ranking does not deviate from the safety baseline.

[0064] 2. Generate a candidate action pool For local repair sections, ranging in length from 200m to 2000m, the system automatically generates all possible repair actions, including tower relocation, tower insertion, tower height increase, and fragment rearrangement, and calculates the three main target performance indicators for each action. For example: Tower relocation action: Calculate the actual marginal cost of each candidate tower location within the search range of ±20m~50m, such as the incremental cost of foundation construction, the horizontal disturbance score, such as the distance from the original tower location, and the construction feasibility, such as whether the slope meets the standard. Tower insertion action: Calculate the cumulative cost of different tower insertion locations, span disturbances such as whether the difference between adjacent spans is ≤30%, and construction accessibility through dynamic programming.

[0065] 3. Screen non-dominated solutions and construct the Pareto front. Iteratively filter the candidate action pool based on "Pareto dominance": ① Compare the three major indicators of each action and eliminate all controlled solutions. For example, if the cost of action X is higher than that of action Y, the disturbance is greater than that of Y, and the feasibility is lower than that of Y, then X is eliminated. ② The remaining non-dominated solutions constitute the Pareto front. For example, the Pareto front of a certain repair segment may contain 3 actions: Action A: Lowest cost (100,000 yuan), but moderate disturbance (15m offset from the original tower location), feasibility meets the standard; Action B: Moderate cost (150,000 yuan), minimal disturbance (5m offset), excellent feasibility; Action C: Higher cost (200,000 yuan), less disturbance (8m offset), better feasibility (suitable for complex terrain sections).

[0066] 4. Sort by strategy weight to determine action priority. Non-dominated solutions on the Pareto front need to be prioritized according to engineering requirements. This invention provides a dynamic weight adjustment mechanism: If it is in the design stage and the line level is 1000kV UHV, the construction feasibility weight can be increased (e.g., weight 0.4) and the cost weight can be reduced (e.g., weight 0.3). Action B, which has less disturbance and better feasibility, should be given priority. If the line is under construction and is 35kV low voltage, the cost weight can be increased to 0.5, and action A, which has the lowest cost, can be selected first. The weight adjustment methods include three types: predefined strategy table, real-time dynamic update, and adaptive optimization, to ensure that the sorting adapts to different engineering scenarios.

[0067] If, after the best action in the Pareto front sort is executed, a violation of hard constraints is found through data snapshot rollback, then that action is removed from the front and the next best action is reselected. In this invention, the multi-objective optimization weights—namely, the weights for safety, true marginal cost, local disturbance, and construction feasibility—can be adjusted in real time according to the project stage, line grade, and on-site strategy. These multi-objective optimization weights can be adjusted in real time based on the project stage, line grade, and on-site strategy, including setting weights for safety, true marginal cost, local disturbance, and construction feasibility separately, and dynamically updating them during the calculation of local repair actions. This ensures that the selection of candidate actions balances safety and economy, and adapts to different construction environments and project stages.

[0068] The weight adjustment methods include three approaches: predefined strategy tables, real-time dynamic updates, and adaptive optimization. Predefined strategy tables establish weight tables based on project stage, line grade, and site type, automatically calling the corresponding weights before each local repair action calculation. Real-time dynamic updates adjust weights in real time based on verification results, construction constraints, or cost changes, linking with the local repair strategy tree and influencing the selection order of action nodes. Adaptive optimization uses a feedback mechanism to automatically fine-tune weights based on repair results such as verification pass rate and cost increments, ensuring the optimality and stability of local repair action selection.

[0069] The data snapshot rollback mechanism of this invention follows a logic of first storing the snapshot as a baseline → local simulation trial calculation → applying according to the rules → rolling back and optimizing for non-compliance. Specific steps include: ① Preliminary preparation: Generate data snapshot Define the snapshot scope: This includes the tower section affected by localized repairs and its adjacent impact area, including potentially affected neighboring towers. Collect core parameters, fully recording tower coordinates, tower type parameters including model, elevation, turning angle level, span information, sag curves under different meteorological conditions, four hard constraints (sag, ground clearance, crossing, wind deflection check indicators), and local cost surface data, forming the original scheme baseline snapshot.

[0070] ② Local trial calculation: Independent simulation repair based on snapshot Create a simulation environment: Based on the snapshot generated in step ①, build a local independent computing environment that is physically isolated from the global tower ranking model.

[0071] Execute repair actions: In the simulation environment, candidate repair actions such as tower relocation, tower height increase, and tower insertion are generated for local repair sections, and action simulation, hard constraint verification, and real marginal cost calculation are performed.

[0072] ③ Compliance assessment: Verify the feasibility of the remediation actions. Judgment criteria: Three conditions must be met simultaneously: hard constraints including sag, clearance, cross-span, and wind deflection must all meet the standards; the actual marginal cost must not exceed a preset threshold, which can be set as 15% of the total cost of the repair section; and construction feasibility must be compliant, such as the slope of the tower relocation location being ≤25°.

[0073] Branching path: Compliance: Proceed to step ④ to apply the remediation actions to the global model; Non-compliant: Proceed to step ⑤ to trigger the rollback process.

[0074] ④ Compliant Application: Synchronously update the global model Global synchronization: Synchronize the repair actions and adjusted parameters that have passed the local simulation to the global tower ranking model in one go.

[0075] Log recording: Detailed records of repair action type, target tower location, adjusted parameters, cost increment, and verification results, forming a traceable execution log.

[0076] Process transition: Proceed to the next round of global optimization or local iterative verification until all constraints are met.

[0077] ⑤ Non-compliant rollback: Restore to the original state and optimize. Trigger rollback: Call the original data snapshot generated in step ① to restore all parameters such as tower position, span, and sag of the affected local segment and adjacent affected domain to the state before the repair action was executed, and clear the potential impact of invalid actions on the global model.

[0078] Action classification and processing: Invalid actions: Actions marked as blacklisted to prevent the generation of the same invalid actions in the future; Optimizable actions: Adjust parameters such as tower shift offset and tower insertion spacing combination, and re-incorporate them into the next round of candidate repair action generation strategy.

[0079] Log retention: Records the reasons for rollback triggers, such as constraint violations, cost overruns, snapshot versions, and rollback times, providing data support for the dynamic updating of the strategy tree for remediation actions.

[0080] ⑥ Iterative Loop: Re-enter the trial calculation process After the rollback is complete, return to step ② and re-execute the trial calculation and compliance judgment in the simulation environment built from the original snapshot based on the updated candidate actions or parameters until a remediation plan that meets the requirements is generated.

[0081] The following table compares the parameters and application scenarios of the data snapshot rollback mechanism:

[0082] In step (4), the output includes repair action instructions, changes to the tower location range, cost predictions, and verification results. The output repair action instructions should include the following: Types: Tower relocation, tower insertion, height increase, fragment rearrangement, tower upgrade; Target tower number: The tower position that indicates the action taken; Position adjustment amount: horizontal and vertical fine-tuning values; Tower type adjustment information: Tower type upgrade or gear spacing adjustment; Action execution order: If there are multiple repair actions, the output order and dependencies.

[0083] The output of the changed tower location range should include the following: Affected tower segment: starting tower number – ending tower number; Original tower location coordinates and adjusted coordinates; Description of the adjustments and disturbances relative to the original plan; The output cost prediction should include the following: The true marginal cost of a local segment; The incremental cost of the segment is the cost incurred due to tower relocation, tower insertion, height increase, or tower type upgrade. Incremental node costs and changes in single-tower construction costs.

[0084] The output verification result description should include the following: Each repair action corresponds to a verification indicator; Verification status marker; Provide detailed data tables or visual reports to support construction or review.

[0085] In this invention, when the line crosses railways, buildings, rivers and other important features, the clearance, wind deflection and safety gap check thresholds are dynamically increased by introducing corresponding exclusive penalty coefficients, so that the assessment results can reflect the high-risk situation of the actual project and improve the safety margin.

[0086] When crossing sensitive features, this invention applies a penalty amplification to the minimum safe clearance distance and the crossing safety gap based on the target type and risk level: Penalty coefficient expression:

[0087] in: Indicates the revised standard limit. This represents the penalty coefficient, which is dynamically adjusted. By accumulating data on the correlation between penalty coefficients and safety accident rates for similar projects, the coefficient is iteratively optimized. If a certain type of feature, such as crossing a highway, still experiences two instances of insufficient clearance after applying a penalty coefficient of 1.5, the coefficient for that scenario is increased to 1.7.

[0088] In this invention, the cost calculation module comprehensively considers the costs of foundation construction, tower materials, crossing engineering, and the costs of demolishing and renovating existing towers. It also performs quantitative analysis on the increase in engineering costs caused by local adjustments, so as to achieve economic evaluation and optimization decision-making for repair actions.

[0089] Marginal cost quantification methods: , And incorporate it into a multi-objective optimization framework: , This invention combines end-to-end engineering cost assessment with real-time feedback and repair decision-making, enabling dynamic correlation between DEM, tower type library, and cross-span. The cost model is directly embedded into the ranking optimization iteration, forming a safety-economic closed-loop control.

[0090] The following is a detailed explanation of how dynamic programming is used to calculate the cumulative true marginal cost: The dynamic programming state includes setting the set of candidate tower locations along the cumulative distance of the line within the local repair section as P = {p1, p2, ..., p...}. n}, where n is the number of candidate locations, determined based on the repair section length of 200m~2000m and the terrain discretization accuracy, the state definition of dynamic programming is: , Where i represents the i-th candidate tower placement position (1≤i≤n); j represents the type of adjacent span combination formed after tower installation (j=1,2,...,k, k is the number of compliant span combinations, which must meet the upper and lower span requirements in the line design specifications). C represents the cumulative real marginal cost from the start of the repair segment to the i-th candidate position, using the j-th range combination; S is the constraint satisfaction status flag. S=1 indicates that the sag, clearance, wind deflection, and cross-span constraints are all satisfied in this state; S=0 indicates that there is a constraint violation.

[0091] Phase division: Phase division is based on the cumulative distance order of candidate tower placement positions, with each candidate position corresponding to a decision phase. The phase number is consistent with the candidate position number, i.e., phase 1 corresponds to candidate position p1, phase 2 corresponds to candidate position p2, and so on, until phase n corresponds to candidate position p1. n .

[0092] Decision variables: The decision variables for each stage are "whether to place a tower at the current candidate position" and "the type of span combination to use after placing the tower", namely: Decision Option 1: Do not place the tower at the i-th candidate position, and directly proceed to the next stage i+1; Decision Option 2: Insert the tower at the i-th candidate position and select the j-th span combination. The selected j-th span combination must satisfy the condition that the sum of adjacent spans equals the length of the corresponding interval of the repair segment.

[0093] Cumulative true marginal cost calculation rules: Regarding cost composition and quantification standards, the cumulative true marginal cost C is calculated using the following formula: , The incremental cost of the project along the route in stage m includes the increased cost of conductor usage due to changes in span length and the cost of deploying construction machinery, calculated as follows: = k1 xΔL + k2 xΔ Calculate, where ΔL is the change in gear ratio, Δ K1 represents the construction difficulty coefficient caused by the sag variation, and K2 represents the industry benchmark unit price coefficient. The incremental cost of the tower location node in stage m includes the cost of tower materials, foundation construction costs, and auxiliary support costs. , The cost weighting coefficient is dynamically adjusted according to the project stage and line grade, such as the construction stage. =0.4、 =0.6.

[0094] Regarding the adjustment of constraint-related costs: If a decision at a certain stage results in a constraint violation S=0, the penalty cost C for the constraint violation must be added. 惩罚 ,Right now: C 调整 = C + x C 惩罚 , As a penalty weight, ≥1, in scenarios involving sensitive ground features =2; C 惩罚 = K 惩罚 X C 基准 K 惩罚 For the cross-specific penalty coefficient, C 基准 The routine rectification costs for violations of this constraint.

[0095] Regarding the state transition equation for the no-tower-insertion decision in this invention, if no tower is inserted in the i-th stage, the state transition is as follows: , Where j' is the gear spacing combination type corresponding to the (i+1)th stage, which needs to be continuously adapted to the gear spacing combination of the ith stage. If the constraints in the previous stage satisfy state S=0, then the state flag remains S=0 after the transition.

[0096] Regarding the state transition equation for tower insertion decision in this invention, if tower insertion is performed in the i-th stage and the j-th span combination is selected, the state transition is as follows: , Let j0 be the cumulative cost of the optimal state in the (i-1)th stage, and j0 be the optimal range combination type in the (i-1)th stage. Ci Let be the actual marginal cost of inserting the tower in stage i; S i The constraint satisfaction state after tower insertion in stage i is determined after checking sag, clearance, wind deflection, and cross-span.

[0097] Regarding the boundary conditions and initial state of this invention, the initial state corresponding to the starting point of the repair segment is: , Where, d p [0][0] indicates the initial state when no candidate position has been entered and there is no gear combination; the initial cumulative cost is 0, the constraint satisfaction state is marked as 1, and there is no constraint violation at the default starting point.

[0098] Boundary constraints include the following three points: First, candidate tower locations must meet construction feasibility requirements. Locations with a slope exceeding 25° are directly marked as infeasible and not included in the state calculation. Second point: The span combination must meet the line design specifications, and the difference between adjacent spans must not exceed 30%; otherwise, the combination type j is invalid. Thirdly, the cumulative true marginal cost must not exceed a preset threshold, which is set at 15% of the total cost of the repair section. If the cost exceeds the threshold, the section will be pruned and will not be included in the subsequent transfer.

[0099] Regarding the optimal solution backtracking process of this invention, the backtracking trigger condition is as follows: When the calculation reaches the last candidate position p n After the corresponding stage n, the backtracking process is triggered, prioritizing the selection of states that meet the following conditions: 1. The constraint satisfies state S=1; 2. The cumulative true marginal cost C is minimized.

[0100] The optimal solution backtracking steps of this invention include: 1. From the optimal state dp[n][j] in stage n opt [Start, trace back to stage 1;] 2. Record the decision results at each stage, including whether to insert towers and the type of range combination; 3. If a constraint violation state S=0 is encountered during the backtracking process, the process jumps to the suboptimal state of that stage and re-backtracks; 4. Output the complete decision path, including the optimal tower placement position p. opt And the corresponding adjacent gear spacing combinations.

[0101] Result verification: The optimal solution obtained through backtracking needs to be checked against constraints again to confirm that the sag, clearance, wind deflection and crossing indicators all meet the requirements. At the same time, the final cumulative real marginal cost is calculated as the final evaluation result of the tower insertion repair action.

[0102] Repair actions are prioritized based on safety over economy. This means that defects in the verification results that pose safety hazards (including but not limited to insufficient clearance, excessive wind deflection, and failure to meet cross-span safety distance requirements) are given the highest weight constraint response. When multiple repair actions can eliminate safety hazards, the system further prioritizes them based on factors such as cost increment, construction accessibility, and ease of operation and maintenance. Through a hierarchical decision-making mechanism, the safety priority is not weakened by local cost-optimal solutions, thereby ensuring that the tower placement scheme meets national and industry safety standards and is feasible for engineering implementation.

[0103] During local optimization, the system can automatically cache data snapshots of the original tower placement scheme, including tower coordinates, tower type parameters, span organization, verification index results, and cost estimation information, to support designers in manual comparison, difference review, and engineering decision verification. If the repair action causes the local scheme to lose its implementation advantages in terms of economy, schedule, or construction, the system allows one-click rollback to the previous version scheme and retains the entire process record of the repair attempt for subsequent intelligent learning and strategy optimization, thereby improving the controllability of the placement repair process and the reliability of the engineering design.

[0104] After performing any repair action and completing the iterative update, the system automatically reconstructs the local cost surface of the affected line section. The updated content includes the cost of the tower foundation, the cost of tower materials, the additional costs of crossing and construction, and the cost of adjusting the tower height due to changes in sag. This reflects in real time the impact of the latest tower arrangement and changes in verification constraints on the project's economic efficiency. The reconstructed cost surface is used to guide the priority ranking and feasibility assessment of subsequent iterative actions, ensuring that the optimization process maintains optimal economic benefits while meeting safety constraints.

[0105] The system user interface supports the visualization of local line sections, including the distribution of conductor sag curves, weak points in ground clearance, locations of safety wind deviation risks, and engineering risk characteristics such as crossing sensitive areas. It also displays the changes in the above indicators of the ranking scheme before and after the repair. Users can zoom in, select local areas, and mark risks in the interface to enable designers to quickly identify design weaknesses and confirm the engineering decision on the effect of repair actions.

[0106] This invention enables parallel computing to accelerate repair efficiency for multiple local segments that do not meet constraints: For multiple local segments that do not meet hard constraints during the placement of transmission line towers, a parallel computing mechanism is provided to improve local repair efficiency. Each local segment is geographically or geographically independent, allowing for the simultaneous generation, verification, and cost calculation of repair actions, thereby shortening the overall optimization iteration time. The specific processing flow is shown below.

[0107] The system calculates the influence domain for each segment to be repaired, covering the segment and adjacent towers that may be affected by the repair operation. A conflict graph is constructed based on the overlap relationship of the influence domain, and the segments are divided into mutually exclusive parallel batches using graph coloring or the maximum independent set algorithm to ensure that operations between segments within the same batch do not conflict with each other.

[0108] The influence domain of this invention is used to define the range that adjustments to a certain local repair section may affect during the local repair of transmission line towers. Essentially, it is the boundary range associated with geographical space and engineering constraints. That is, the area within the repair section where tower adjustments, such as moving or inserting towers, may affect the surrounding tower positions, spans, conductor sag, and safety verification indicators. It is necessary to clearly define the scope to avoid conflicts between adjustment actions in different repair sections, thus providing a basis for parallel computing and global data consistency control.

[0109] During localized repairs, actions on a single repair section, such as moving or installing towers, not only affect the towers within that section but can also impact the safety constraints of adjacent towers, such as clearance and wind deflection, through span relationships and sag chain reactions. For example, moving tower T-26 to repair section T-25~T-27 will change the spans of T-25~T-26 and T-26~T-27, thereby affecting the conductor sag of these two spans and indirectly impacting the clearance check results of T-24 adjacent to T-25 and T-28 adjacent to T-27.

[0110] The affected area must cover the repair section itself and the adjacent towers and spans that may be indirectly affected by its actions. This ensures that the feasibility of the repair actions is assessed within this area, and avoids local repairs leading to new violations in the surrounding area or conflicts between multiple parallel repair actions.

[0111] The calculation of the influence domain needs to be carried out around three dimensions: geospatial, engineering constraints, and action type, and specifically includes the following three steps: Step 1: Determine the initial boundary based on the local repair segment. The length of the local repair section ranges from 200m to 2000m, dynamically set according to the distribution of violations. The initial range of the influence area is based on the starting and ending tower positions of the repair section, first covering all towers and spans within the section. For example, if the repair section is towers T-23 to T-27, corresponding to a line length of 800m, then the initial boundary of the influence area is the line section from tower T-23 to tower T-27, including five towers and four spans within the section.

[0112] Step 2: Extend to neighboring tower locations according to hard constraint association rules. Adjustments to repair work will affect adjacent tower positions through the chain relationship of span, sag, and safety constraints. The number of adjacent tower positions requiring expansion needs to be determined based on the calculation logic of the four hard constraints, with the core basis being the impact range of span changes on sag. Sag Constraint Correlation: The calculation of conductor sag depends on parameters of a single span, such as span length and tower height. However, changes in sag between adjacent spans can indirectly affect conductor tension at the top of the tower, especially under meteorological conditions such as icing and strong winds. Therefore, it is necessary to include 1-2 towers before and after the repair section in the influence zone. For example, for repair sections T-23 to T-27, the influence zone needs to be extended to T-22 to T-28 to ensure coverage of all tower locations that may be affected by the sag cascading effect.

[0113] Clearance / Intersection / Wind Deflection Constraints: The verification of these three types of constraints depends on the conductor position within the span, such as the lowest point of sag and the lateral offset after wind deflection. If the span changes within the repair section, it will directly change the conductor position within that span. Simultaneously, tower position shifts may affect the distance between the conductors in adjacent spans and ground features / intersection targets. For example, if tower T-26 is located within the repair section, moving tower T-26 will change the lowest point of sag for spans T-25 to T-26, and may also change the wind deflection clearance for spans T-26 to T-27. Therefore, the adjacent towers T-24 and T-28 of T-25 and T-27 need to be included in the influence domain to ensure the verification of the span constraints corresponding to these tower locations.

[0114] Step 3: Dynamically adjust the range based on the type of repair action. Different repair actions have varying degrees of impact on the surrounding environment, requiring further fine-tuning of the influence domain based on the characteristics of the action. The adjustment rules corresponding to different action types are as follows:

[0115] The influence domain is designed to support parallel computing and conflict control. The ultimate purpose of calculating the influence domain is to serve the parallel repair of multiple local segments. 1. Constructing a conflict diagram: The system performs spatial overlay analysis on the influence domain of each repair segment. If the influence domains of two repair segments overlap, such as the influence domains of T-23~T-27 overlapping with the influence domains of T-26~T-30 in T-26~T-27, then it is determined that these two repair segments have action conflicts and need to be assigned to different parallel batches. 2. Parallel batch division: Using graph coloring algorithm or maximum independent set algorithm, repair segments with no overlapping influence domains are divided into the same batch to ensure that repair actions within the same batch do not interfere with each other. For example, the influence domains of T-23~T-27 and T-31~T-35 do not overlap and can be repaired in parallel. 3. Ensure global consistency: Before a repair action is submitted, constraint checks such as sag and clearance only need to be performed within its affected domain. No global verification is required, which improves efficiency and ensures that there are no conflicts in the global data by ensuring that there are no conflicts within the affected domain.

[0116] The influence domain is a local repair impact boundary based on the repair segment, combined with constraint correlation and action characteristics quantification. The calculation logic can be summarized as follows: based on the dynamically set repair segment of 200m~2000m, it is extended to 1~2 adjacent towers before and after according to the chain reaction of hard constraints. Then, the range is fine-tuned by combining the impact degree of tower moving, tower insertion and other actions, and finally forms a boundary covering all possible affected towers and spans, which provides support for the efficiency of parallel repair and the consistency of global data.

[0117] Each parallel task performs repair action calculations on a local data snapshot, including candidate action generation, sag / headroom / wind deflection / crossing verification, and true marginal cost assessment.

[0118] Candidate actions can be evaluated in parallel within a segment, enabling efficient computation of action generation, Pareto filtering, and optimal action selection.

[0119] After the trial calculation is completed, atomic updates are performed using an optimistic concurrent commit mechanism, followed by conflict detection. If a conflict in the affected domain is detected, a rollback or rescheduling is executed to ensure global data consistency. The optimistic concurrent commit mechanism is a technical strategy used in the local repair process of transmission line tower positioning to ensure global data consistency when multiple local segments execute repair actions in parallel. It first attempts to execute the repair independently, then verifies whether there is a conflict. It assumes that the repair actions of each local segment do not interfere with each other, and checks for conflicts only before submitting the final result. This avoids efficiency losses caused by prematurely locking resources, while ensuring the safety and controllability of the global solution through a conflict handling mechanism. The optimistic concurrent commit mechanism is based on a trial-and-error approach, adjusting only in case of conflicts. In scenarios involving parallel repair of multiple local segments, such as simultaneously repairing two independent segments, T-23~T-27 and T-35~T-39, traditional pessimistic concurrency would pre-lock repair segment resources, resulting in only one segment being processed at a time, leading to low efficiency. Optimistic concurrent commit, however, is based on the premise that the influence domains of local segments are relatively independent—that is, the influence domain is the boundary range affected by the repair action. It allows each segment to perform repair trials simultaneously, checking for conflicts such as overlapping influence domains or data overwriting only when the repair results are submitted to the global model. This achieves a balance between efficient parallelism and global consistency. The implementation process of the optimistic concurrent commit mechanism includes the following four steps: Step 1: Independent local trial calculations without interfering with the global process. For each segment to be repaired, a complete repair process is executed on its dedicated local data snapshot, containing data such as tower coordinates, span, and sag curves for that segment and its affected area. Candidate actions are generated, such as tower relocation, tower insertion, Pareto sorting, execution of the optimal action, and constraint verification, resulting in the repaired local solution for that segment. The trial calculation process is entirely performed within the snapshot copy, without modifying the original data of the global tower positioning model. Trial calculations for each segment can be performed simultaneously and in parallel, such as T-23~T-27 in batch 1 and T-35~T-39 in batch 2, avoiding resource contention.

[0120] Step 2: Conflict detection before submission. After all parallel segments have completed the trial calculation, the system compares the overlap of the influence domains of the local solutions after each segment is repaired. The influence domain is the range that the repair action may affect, including the segment and 1-2 adjacent base towers. If the influence domains of two segments do not overlap, such as the influence domains of T-23~T-27 being T-22~T-28, and the influence domains of T-35~T-39 being T-34~T-40, then no conflict is determined. If the influence domains overlap, such as the influence domains of T-26~T-30 and T-28~T-32 both including T-28~T-30, then a conflict is determined.

[0121] In addition to the overlap of the affected areas, data consistency also needs to be verified. For example, after the repair of section A, the T-28 tower position is moved 10m to the east, and after the repair of section B, the T-28 tower position needs to be moved 5m to the west. The two adjustments to the same tower position are contradictory, and even if the affected areas do not completely overlap, they are still judged as a conflict.

[0122] Step 3: If there are no conflicts, submit and update the global system synchronously. If there are no conflicts in the repair solutions of all parallel segments, the system will synchronously update the repaired local solutions of each segment to the global tower positioning model at one time. The update includes key data such as tower coordinates, span combination, sag curve, and cost surface. At the same time, the system records the repair log action type, cost increment, and verification results of each segment for subsequent traceability.

[0123] Step 4: Rollback and Rescheduling in Case of Conflicts. Rollback Operation: If a conflict is detected, such as overlapping influence domains of two segments or contradictory tower position adjustments, the system immediately terminates the submission process and rolls back all segments involved in the conflict to their original snapshot state before the trial calculation. This means canceling the repair trial calculation results of these segments and restoring their tower positions, spans, and other data to their initial state to prevent erroneous data from entering the global model. For the conflicted segments after rollback, the parallel batches are re-divided. For example, the original T-26~T-30 and T-28~T-32 in the same batch are split into two batches. T-26~T-30 is repaired first, and then T-28~T-32 is executed after submission. Alternatively, the repair action parameters are adjusted, such as modifying the tower relocation range of one of the segments. The trial calculation and conflict detection are restarted until there are no conflicts before submission.

[0124] The system records a snapshot of the original data for each local segment and the repair attempt process. If the commit fails or is blocked by conflicts, it can automatically roll back to the original state and retain logs for manual review or subsequent strategy learning.

[0125] Through this parallel computing mechanism, the present invention can significantly improve local repair efficiency, shorten the tower placement optimization iteration time, ensure that local and global safety and economic constraints are met, and provide scalable and reliable intelligent optimization solutions for complex lines and large-scale engineering applications.

[0126] The system of this invention supports the comparison of sag curves before and after changing the tower position, which is used to demonstrate the optimization effect.

[0127] The repair system of this invention applies each local repair action on a local data snapshot and performs constraint checks and feasibility verification before submitting it to the global tower placement model. When the verification results show that the repair action violates hard constraints or fails to meet safety / economic requirements, the system automatically cancels the repair action, retaining only the original tower placement scheme to ensure that infeasible solutions do not propagate to the global placement. Simultaneously, the system records the rollback operation and the evaluation results of the corresponding candidate actions for subsequent action selection, strategy optimization, and manual review. In specific implementation, the rollback process includes the following steps: Before performing repair actions, a data snapshot is generated for the affected local section and its adjacent affected area. The snapshot includes tower location coordinates, tower type parameters, span information, sag curve, verification indicators, and local cost surface. Simulate, verify, and calculate costs for candidate actions on a local snapshot, and select the optimal repair action to attempt application; If the verification or constraint validation fails, the system uses a snapshot to restore the local segment to its original state and removes the impact of the attempted repair actions on the global model. After the rollback operation is completed, the system can mark the candidate action as infeasible, or use its parameters to adjust the generation strategy for the next round of repair actions.

[0128] Through the above mechanism, the system ensures that the infeasibility of each repair action during the local optimization process will not spread to the global ranking, thus ensuring that the global tower ranking scheme always meets the constraints of safety, economy and engineering feasibility.

[0129] The repair model used in this invention is applicable to voltage levels from 35kV to 1000kV and different meteorological zones.

[0130] The local repair system used in this invention can be automatically coupled with the global tower positioning optimization module to form a fully intelligent closed-loop positioning system, thereby achieving integrated closed-loop control from initial positioning generation, global optimization, local verification and repair to the final positioning scheme. The specific implementation process is as follows... Figure 6 This includes the following steps: Global Ranking Generation: The system generates an initial global ranking scheme based on the line start and end points, tower type library, and engineering constraints, including tower coordinates, tower type, and span information; Global optimization: The global optimization module optimizes the initial scheme based on the overall cost surface, engineering constraints, and construction feasibility to obtain a global candidate ranking scheme; Local verification and repair: The local repair module receives the global optimization results and performs repair action generation, Pareto filtering, real marginal cost evaluation, and optimal action application on local segments that do not meet the constraints of sag, clearance, wind deflection, and cross-span. During the repair process, data snapshots, parallel computing, and rollback mechanisms are supported to ensure that infeasible solutions do not spread to the global solution. Results feedback and closed-loop iteration: After the local repair is completed, the repair results, cost surface update and verification feedback information are automatically transmitted back to the global optimization module as input for the next round of global optimization or local repair strategy adjustment, so as to realize the iterative coupling between local and global. Closed-loop mechanism: The system iterates continuously until all towers along the entire line meet safety, economic and construction constraints, and generates the final overall ranking plan. At the same time, it retains local repair history, iteration logs and cost assessment data for engineering decision review and intelligent strategy learning.

[0131] This invention generates an initial overall tower configuration through a global ranking generation module, followed by overall cost and constraint optimization through a global optimization module. A local verification and repair module receives the global optimization results and performs candidate repair actions, Pareto filtering, and real marginal cost assessment for local segments that do not meet constraints. It also supports data snapshots, parallel computing, and rollback mechanisms to ensure that infeasible solutions do not propagate globally. Local repair results and cost surface updates are automatically fed back to the global module, achieving closed-loop iteration until all towers along the entire line meet safety, economic, and construction constraints, generating a final ranking scheme and forming an intelligent closed-loop ranking system for the entire line.

[0132] The above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for local adjustment of transmission line tower alignment, characterized in that, The method comprises the following steps: 1) fitting the conductor sag of the generated tower arrangement scheme, obtaining the position distribution of the conductor along the line, and then comprehensively judging the sag checking, the ground clearance checking, the crossing checking and the wind deviation checking, wherein the comprehensive judgment needs to first confirm whether the single checking dimension meets the respective safety standard, and then confirm whether the multi-dimensional constraint meets the standard at the same time, if any dimension does not meet the standard, it is determined as a violation, triggering the subsequent local repair process, if all dimensions meet the standard, it is determined as qualified and no repair is needed; 2) identifying the tower position that does not meet the hard constraint according to the checking result of step 1), determining the local repair section by the tower position and the adjacent tower positions before and after the tower position, wherein the hard constraint comprises the sag checking, the ground clearance checking, the crossing checking and the wind deviation checking, and the length of the local repair section is set according to the distribution of the violation constraint; 3) generating multiple candidate repair actions for the local repair section determined in step 2), including but not limited to tower moving, tower height increasing, tower inserting and / or local segment rearrangement, and each candidate repair action corresponds to at least one adjustment parameter; 4) constructing a multi-objective local optimization model, quantifying the repair cost of different candidate repair actions based on the real marginal cost on the premise of meeting the hard constraint, and then sorting the candidate repair actions and outputting the sorting result, wherein the real marginal cost is determined by the weighted superposition of the line engineering cost increment and the tower position node cost increment; 5) performing the repair action according to the sorting priority output in step 4), and re-executing steps 1) to 4) on the adjusted local scheme until the local scheme meets the hard constraint.

2. The method for local adjustment of transmission line tower alignment according to claim 1, wherein: In the step 1), the sag checking comprises judging whether the geometric shape of the conductor under different meteorological conditions is reasonable, whether the conductor tension and stress level is within the safe and controllable range, whether the sag size meets the design threshold, and whether the sag data meets the calculation requirements of subsequent clearance checking and wind deviation checking; The ground clearance checking is first based on the parabolic approximation to calculate the sag, and when the minimum clearance is less than the configured threshold, it is switched to the catenary explicit solver to obtain the accurate lowest point coordinates, and then the catenary or the segmented fitting model is used to obtain the conductor sag curve, the ground DEM or the ground object model is extracted within the continuous span range to search for the sag lowest point, and the difference between the conductor height at the position x on the sag curve and the highest point elevation of the corresponding position ground or ground object is calculated as the clearance value, if the clearance value is less than the set standard, the checking does not meet the standard; The crossing checking comprises judging whether the vertical clearance and the horizontal offset distance between the conductor lowest point and the highest point of the crossing object meet the standard, and additionally judging whether the electric field influence is compliant in the high-risk scenario; The wind deviation checking is based on the maximum wind deviation angle allowed by the selected tower insulator string as the control index, and the transverse offset curve of the conductor under the action of wind load is calculated by combining the reference wind speed, air density, terrain roughness and wind load increase coefficient of the meteorological area where the line is located, and it is judged whether the safety clearance between the offset conductor and the ground object, the crossing target and the adjacent phase conductor meets the standard, and whether the insulator string swing angle exceeds the maximum allowed swing angle.

3. The method for local adjustment of transmission line tower alignment according to claim 1, wherein: The tower moving action in step 3) searches the DEM elevation and slope data within a range of ±20 m to 50 m around the center position of the target tower site, combines the construction feasibility constraints and the sag, clearance, wind deviation and intersection crossing hard constraints, calculates the real marginal cost and horizontal disturbance score of each candidate tower site, and selects the candidate tower site with the highest comprehensive score as the final tower moving position; The tower inserting action calculates the cumulative real marginal cost of different candidate tower sites in the local repair section under the constraints of the span, sag, clearance, wind deviation and intersection crossing hard constraints, and generates the optimal tower inserting position and adjacent span combination through backtracking; The local segment rearrangement action combines and transforms the local tower site segments containing 3 to 5 basic towers, changes the relative positions and span combinations of adjacent towers, performs the sag, clearance, wind deviation, intersection crossing and construction feasibility check, and selects the optimal tower site combination with the minimum real marginal cost from the candidate combinations that meet the hard constraints; When the tower height increasing action automatically selects a higher tower type or a special corner tower type within a set range, it checks the corner level, maximum horizontal span and height conditions in real time, combines the local sag, clearance, wind deviation and intersection crossing check results, and selects the optimal tower type with the minimum real marginal cost that meets the hard constraints.

4. The method for local adjustment of transmission line tower alignment of claim 1, wherein: In step 3), when generating candidate repair actions, a repair action generation strategy tree is constructed based on the check feedback results, the repair action generation strategy tree includes action nodes, decision nodes and terminal nodes, the action nodes are tower moving, tower inserting and tower height increasing, the decision nodes are constraint satisfaction judgments, and the terminal nodes are feasible scheme outputs; In the branch design of the repair action generation strategy tree, the action priority is predefined, each node is bound with three types of node indicators, namely real marginal cost, construction feasibility and local disturbance, and after each local action is executed, the repair action generation strategy tree is dynamically updated based on the check results.

5. The method for local adjustment of transmission line tower alignment of claim 1, wherein, The repair action and check feedback generated strategy tree in step 3) includes: Based on the comprehensive check results of step 1), the violation types, constraint compliance states and local repair segment parameters are extracted to determine the initial input conditions for constructing the strategy tree; The node composition of the strategy tree is defined, wherein the action nodes correspond to the three types of core repair actions of tower moving, tower inserting and tower height increasing, the decision nodes are hard constraint satisfaction judgments, the terminal nodes are feasible repair scheme outputs, and the mapping relationship between the nodes and the repair actions is established; The branch structure of the strategy tree is designed, the action priority of safety over economy is preset in the branch, three types of indicators, namely real marginal cost, construction feasibility and local disturbance, are bound for each action node, and the indicator quantification rules are clearly defined; The branch traversal of the strategy tree is triggered, the initial input conditions are matched with the action nodes to generate corresponding candidate repair actions, and the repair actions are simulated; Through the decision node, it is judged whether the simulated execution meets the hard constraints of sag, ground clearance, intersection crossing and wind deviation: if yes, the repair action and parameters are output to the terminal node to form a feasible scheme; if not, the indicator weights of the action nodes are adjusted based on the new check feedback results, and the candidate repair actions are matched again; After each local repair action is executed, the node data, branch priority and action parameters of the strategy tree are updated synchronously, each node saves the original scheme and simulation results, and manual review or automatic rollback is supported.

6. The method for local adjustment of transmission line tower alignment of claim 1, wherein, Step 4) The candidate repair actions are sorted by using Pareto front sorting to screen non-dominated solutions, specifically including the following steps: a) The optimization target and index quantization rules are defined, the optimization target includes economy, engineering stability and construction feasibility, the economy is quantified by the real marginal cost, the engineering stability is quantified by the local disturbance, the construction feasibility is quantified by the terrain adaptability and construction period adaptability, the safety is prior to the economy, and the candidate repair actions that do not meet the hard constraints are directly excluded from the Pareto front; b) All possible candidate repair actions are generated for the local repair section to form a candidate action pool, and the economy, engineering stability and construction feasibility index values of each candidate repair action are calculated; c) The candidate action pool is iteratively screened according to the Pareto dominance relationship, and the dominated solutions are eliminated, and the remaining non-dominated solutions constitute the Pareto front; d) The priority of the non-dominated solutions on the Pareto front is determined according to the engineering requirements.

7. The method for local adjustment of transmission line tower alignment of claim 1, wherein: The sorting results output in step 4) include repair action instructions, tower position range changes, cost predictions and review result explanations; The repair action instructions include action type, target tower number, position adjustment amount, tower type adjustment information and action execution sequence; The tower position range changes include affected tower position section, original and adjusted coordinates, adjustment amplitude and disturbance description; The cost prediction includes local section real marginal cost, section cost increment and node cost increment; The review result explanation includes review indicators corresponding to each repair action, review status markers, and provides detailed data tables or visual reports.

8. The method for local adjustment of transmission line tower alignment of claim 1, wherein, When there are multiple local sections that do not meet the constraints, parallel computing is enabled to speed up the repair efficiency, specifically including the following steps: i) Calculate the influence domain for each repair section, which includes the repair section and the adjacent towers that may be affected by the repair action, the influence domain is calculated based on the initial boundary of the local repair section, then expanded to the adjacent towers, and finally adjusted dynamically according to the repair action type; ii) Build a conflict graph based on the overlapping relationship of the influence domains of each repair section, and divide the repair sections into mutually exclusive parallel batches by graph coloring or maximum independent set algorithm; iii) Each repair section in the parallel batch performs repair action trial on the local data snapshot, including candidate action generation, sag / spacing / wind deviation / crossing clearance check and real marginal cost evaluation, and parallel evaluation of candidate repair actions within the section; iv) After completing the trial, update and detect conflicts through an optimistic concurrent commit mechanism, if a conflict in the influence domain is detected, rollback or rescheduling is performed to ensure global data consistency.

9. The method for local adjustment of transmission line tower alignment according to claim 8, wherein, In step iv), the optimistic concurrent commit mechanism includes the following steps: Each repair section performs a complete repair process on a dedicated local data snapshot to generate a repaired local scheme, and the trial process does not modify the original data of the global tower and line arrangement model; After all parallel segments complete the trial, the influence domain overlap and data consistency of each segment after repair are compared to determine whether there is a conflict; If there is no conflict, the local scheme after repair of each segment is updated to the global tower ranking model, and the repair log is recorded; If there is a conflict, the submission process is terminated, the segments involved in the conflict are rolled back to the original snapshot state before the trial, the parallel batches are re-divided or the repair action parameters are adjusted, the trial and conflict detection are restarted until there is no conflict and then submitted.

10. The method for local adjustment of transmission line tower alignment of claim 1, wherein, When there are multiple local segments that do not meet the hard constraints, the repair efficiency is accelerated by parallel computing combined with an optimistic concurrent submission mechanism, which includes: For each segment to be repaired, the initial boundary is determined based on the segment, then it is extended to the adjacent tower position according to the hard constraint association rule, and finally the range is dynamically adjusted according to the repair action type to obtain the influence domain covering the segment to be repaired and the adjacent tower position that may be affected; Based on the overlap relationship of the influence domains of each segment to be repaired, a conflict graph is constructed, and through graph coloring or maximum independent set algorithm, the segments to be repaired without conflict are divided into mutually exclusive parallel batches to ensure that the operations between segments in the same batch do not interfere with each other; Each segment to be repaired in each parallel batch independently executes the repair action trial on the exclusive local data snapshot, including candidate repair action generation, sag / spacing / wind deviation / crossing clearance check and real marginal cost evaluation, and parallel evaluation of candidate repair actions within the segment; After all parallel batches complete the trial, the influence domain overlap and data consistency of each segment after repair are compared to determine whether there is a conflict; If there is no conflict, the local scheme after repair of each segment is updated to the global tower ranking model, and the repair log is recorded; if there is a conflict, the submission process is terminated, the segments involved in the conflict are rolled back to the original snapshot state before the trial, the parallel batches are re-divided or the repair action parameters are adjusted, the trial and conflict detection are restarted until there is no conflict and then submitted.

11. The method for local adjustment of transmission line tower alignment of claim 1, wherein, Before performing the repair action, a data snapshot is generated for the affected local segment and its adjacent influence domain, including tower position coordinates, tower type parameters, span information, sag curve, checking index and local cost surface; On the local snapshot, candidate action simulation, checking and cost calculation are performed, the optimal repair action is selected and attempted to apply, if the checking or constraint verification fails, the local segment is restored to the original state using the data snapshot, the influence of the attempted repair action on the global model is removed, and the candidate action is marked as infeasible or its parameters are adjusted for the next round of repair action generation strategy.

12. The method for local adjustment of transmission line tower alignment of claim 1, wherein, The data snapshot rollback mechanism is used in the process of performing the repair action to ensure that the repair action meets the hard constraints, which includes the following steps: Before starting the local repair process, a complete data snapshot is generated for the affected local segment and its adjacent influence domain, including tower position coordinates, tower type parameters, span information, sag curve, checking index and local cost surface, and the original scheme benchmark is retained; Based on the data snapshot, a local simulation environment is created to perform candidate repair action simulation, sag / spacing / wind deviation / crossing clearance check and real marginal cost calculation without modifying the original data of the global tower ranking model. After the simulation is completed, it is determined whether the repair action meets the hard constraints and is in line with the requirements of economy and construction feasibility: if it meets the requirements, the repair action is applied to the global model; if it does not meet the requirements, a rollback process is triggered; When the rollback process is executed, the generated data snapshot is called to restore all parameters of the affected local section and the adjacent affected domain to the original state before the repair action is executed, and the potential impact of the invalid repair action on the global model is cleared; After the rollback is completed, the candidate repair action is marked as infeasible, or after the parameters are adjusted based on the review feedback, it is re-included in the next round of repair action generation strategy, and the rollback log and candidate action evaluation results are recorded for subsequent strategy optimization and manual review.

13. A system for implementing a local adjustment system for tower arrangement of a power transmission line, the system comprising: a global arrangement generation module for generating an initial global arrangement scheme according to line start and end points, tower type library and engineering constraints; a global optimization module for optimizing the initial global arrangement scheme based on a line-wide cost surface, engineering constraints and construction feasibility to obtain a global candidate arrangement scheme; a local repair module for receiving the global candidate arrangement scheme, performing repair action generation, Pareto screening, real marginal cost evaluation and optimal action application on local sections that do not meet the constraints; After the local repair is completed, the repair results, cost surface update and review feedback information are transmitted back to the global optimization module as input for the next round of global optimization or local repair strategy adjustment.

Images