Power transmission and transformation project management and control method and system based on regional geological characteristics

Abstract

This invention discloses a method and system for power transmission and transformation engineering management based on regional geological characteristics, belonging to the field of power system engineering technology. The method includes the following steps: acquiring power grid and geological data of the target area, constructing a tower spatial map and extracting node features; combining the tower spatial map and node features, generating a set of boundary constraint corridors using a strategic corridor identification method, and constructing a comprehensive cost function within each boundary constraint corridor; based on the boundary constraint corridors, minimizing the comprehensive cost function as the optimization objective, obtaining the target tower layout sequence through the collaborative iteration of Monte Carlo tree search and a physically guided neural network. This invention addresses the problems of low optimization efficiency and decoupling between boundary constraints and line topology in power transmission line design.

Description

Technical Field

[0001] This invention relates to the field of power system engineering technology, and more specifically, to a method and system for the management and control of power transmission and transformation projects based on regional geological characteristics. Background Technology

[0002] As a core component of energy infrastructure, power transmission and transformation projects are crucial for ensuring stable power delivery and economic development. With the continuous expansion of power grid coverage, construction is increasingly extending to areas with complex geographical conditions and high environmental sensitivity. The site selection, design, construction, and operation and maintenance of power transmission and transformation projects are facing increasingly severe geological challenges. Therefore, resolving the growing contradiction between traditional power transmission and transformation project management methods and complex geological environments, and constructing an integrated, intelligent, full life-cycle geologically adaptable management system, has become a vital and urgent industry-wide issue.

[0003] For example, the invention patent announcement CN120561624A discloses an automated adjustment method and system for geological zoning of power transmission and transformation projects. This method includes the following steps: acquiring multi-source geological data and performing data normalization preprocessing; obtaining a high-dimensional feature data table based on a multi-source Bayesian fusion model; constructing and standardizing features in the high-dimensional feature data table to obtain a high-dimensional geological feature set; compressing the high-dimensional geological feature set based on VAE and constructing a deep clustering model based on GMM unsupervised clustering; controlling the power transmission engineering system to perform adaptive geological zoning through the deep clustering model and generating a path control strategy set based on reinforcement learning. This invention, by calling the deep clustering model and reinforcement learning path optimization method, achieves adaptive intelligent adjustment of geological zoning and closed-loop optimization of path control strategies, solving the problems of strong static nature of geological zoning, lack of real-time feedback mechanisms, and lack of deep coupling between path control and geological structure.

[0004] The above-disclosed technical solutions have at least the following technical problems: It is difficult to achieve a precise balance of life-cycle costs and ensure physical feasibility in complex transmission line design tasks; the output results are limited to geological zoning and cannot be directly mapped to the constraint boundaries of line construction; the training phase is complex and computationally intensive, resulting in low optimization efficiency.

[0005] To address the above problems, this invention proposes a solution. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method and system for the management and control of power transmission and transformation projects based on regional geological characteristics. The present invention solves the problems of low optimization efficiency and decoupling of boundary constraints and line topology in power transmission line design by constructing boundary constraint corridors, constructing a comprehensive cost function, and introducing a collaborative iterative optimization framework of Monte Carlo tree search and physical guided neural network.

[0007] To achieve the above objectives, the present invention provides the following technical solution: The power transmission and transformation project management method based on regional geological characteristics includes the following steps: acquiring power grid and geological data of the target area, constructing a tower spatial map and extracting node features; combining the tower spatial map and node features, generating a set of boundary constraint corridors through the strategic corridor identification method, and constructing a comprehensive cost function within each boundary constraint corridor; based on the boundary constraint corridors, minimizing the comprehensive cost function as the optimization objective, obtaining the target tower layout sequence through the collaborative iteration of Monte Carlo tree search and physical guided neural network.

[0008] In a preferred embodiment, the step of generating a set of boundary-constrained corridors by combining the tower space map and node features and using the strategic corridor identification method specifically involves: generating a recommended corridor width for each node based on node features; calculating the attention weights between nodes using a graph neural network model based on node features and outputting the strategic importance score for each node; and combining the strategic importance score with the recommended corridor width to obtain the set of boundary-constrained corridors.

[0009] In a preferred embodiment, the step of combining strategic importance scores and recommended corridor widths to filter and obtain a set of boundary-constrained corridors specifically involves: filtering nodes whose strategic importance scores are greater than a preset score threshold to obtain a set of high-value nodes; expanding the neighborhood around the high-value nodes based on the recommended corridor width to generate corresponding corridor bands; merging the corridor bands and extracting connected components to obtain a set of boundary-constrained corridors.

[0010] In a preferred embodiment, constructing a comprehensive cost function within each boundary constraint corridor specifically involves: extracting nodal physical quantity data from each boundary constraint corridor, including tower physical quantities, pile foundation physical quantities, conductor physical quantities, and market data; based on the nodal physical quantity data, establishing tower construction cost formulas, foundation construction cost formulas, and line operation cross-loss theory formulas under physical constraints; and summing the tower construction cost formulas, foundation construction cost formulas, and line operation cross-loss theory formulas to obtain the comprehensive cost function.

[0011] In a preferred embodiment, the step of obtaining the target tower layout sequence based on boundary constraint corridors, with minimizing the comprehensive cost function as the optimization objective, through the collaborative iteration of Monte Carlo tree search and physical guidance neural network, specifically involves: pre-training the physical guidance neural network based on historical boundary constraint corridors; generating candidate tower layout sequences through Monte Carlo tree search within each boundary constraint corridor, and inputting the candidate tower layout sequences into the pre-trained physical guidance neural network to obtain the predicted comprehensive cost; using the candidate tower layout sequences as the action space and the predicted comprehensive cost as the state space; under safety constraints, negatively mapping the predicted comprehensive cost as a reward signal and feeding it back to the Monte Carlo tree search process to update the candidate tower layout sequences in the action space, thereby obtaining the target tower layout sequence.

[0012] In a preferred embodiment, the physical guidance neural network includes an input layer, a graph message passing layer, a differentiable physical equation module, and an output layer. The input layer receives node physical quantity data, and the graph message passing layer aggregates information based on the spatial topology of the towers. The differentiable physical equation module includes sag theory formulas, conductor actual length theory formulas, and line operation loss theory formulas. The output layer includes a linear activation function.

[0013] In a preferred embodiment, the pre-training of the physical guidance neural network based on the historical boundary constraint corridor specifically involves: constructing a physical guidance neural network based on the historical boundary constraint corridor, using historical node physical quantity data and topological relationships as input; training the physical guidance neural network with the residuals obtained from the differentiable physical equations as the physical loss function, and outputting predicted node physical quantity data; and calculating the predicted comprehensive cost based on the predicted node physical quantity data through a comprehensive cost function.

[0014] In a preferred embodiment, under safety constraints, the step of negatively mapping the predicted comprehensive cost as a reward signal and feeding it back to the Monte Carlo tree search process to update the candidate tower layout sequence in the action space to obtain the target tower layout sequence specifically involves: inputting the reward signal as the reward value into the Monte Carlo tree search and calculating the comprehensive score of the node using the UCT formula; continuously optimizing the node strategy of the Monte Carlo tree search based on the comprehensive score and the cumulative update of the node visit count and action visit count; and generating the target tower layout sequence that satisfies the minimization of the predicted comprehensive cost after iterative convergence.

[0015] The power transmission and transformation engineering management and control system based on regional geological characteristics includes: a data modeling module, used to acquire regional power grid and geological data, construct a tower spatial map and extract node features; a corridor constraint module, used to combine the tower spatial map and node features, generate a set of boundary constraint corridors through strategic corridor identification, and construct a comprehensive cost function within each boundary constraint corridor; and an iterative optimization module, used to obtain the target tower layout sequence based on the boundary constraint corridors, with minimizing the comprehensive cost function as the optimization objective, through the collaborative iteration of Monte Carlo tree search and physically guided neural network.

[0016] The technical effects and advantages of the power transmission and transformation project management method and system based on regional geological characteristics of this invention are as follows: This invention constructs boundary-constrained corridors to rationally limit the design space of transmission lines, and builds a comprehensive cost function within each feasible corridor. It also introduces Monte Carlo tree search and physically guided neural network for collaborative iteration, achieving joint optimization of tower layout, conductor span, and corridor selection. This method effectively solves the problems of low optimization efficiency, incomplete objective consideration, and decoupling of boundary constraints and line topology in traditional transmission line design, achieving coordinated optimization of construction costs, operating losses, and safety and reliability. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the process for a power transmission and transformation project management method based on regional geological characteristics, provided in an embodiment of the present invention.

[0018] Figure 2 This is a schematic diagram of the power transmission and transformation project management and control system based on regional geological characteristics, provided for an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1, Figure 1 This invention presents a power transmission and transformation project management method based on regional geological characteristics, comprising the following steps: S1. Obtain power grid and geological data for the target area, construct a spatial map of the towers and extract node features; S2, combining the tower space map and node characteristics, generates a set of boundary constraint corridors through the strategic corridor identification method, and constructs a comprehensive cost function within each boundary constraint corridor; S3, based on the boundary-constrained corridor, takes minimizing the comprehensive cost function as the optimization objective and obtains the target tower layout sequence through the coordinated iteration of Monte Carlo tree search and physical-guided neural network.

[0021] This embodiment constructs boundary-constrained corridors to reasonably limit the design space of transmission lines, and builds a comprehensive cost function within each feasible corridor. Simultaneously, it introduces Monte Carlo tree search and physically guided neural network for collaborative iteration, achieving joint optimization of tower layout, conductor span, and corridor selection. Through this method, the present invention effectively solves the problems of low optimization efficiency, incomplete objective consideration, and decoupling of boundary constraints and line topology in traditional transmission line design, achieving synergistic optimization of construction costs, operating losses, and safety and reliability.

[0022] S1: Obtain power grid and geological data for the target area, construct a spatial map of the towers, and extract node features.

[0023] In this embodiment, the target area power grid and geological data acquisition includes power grid topology data, terrain data, geological parameters, land use / obstacles, weather / temperature scenarios, tower type library and cost unit price, and load / power flow scenarios; Project all power grid and geological data within the target area onto a unified coordinate system; Based on the power grid and geological data of the target area, the target area is discretized into a grid or triangular grid to construct a pole and tower spatial map. The spatial map uses poles and towers as nodes and the conductors between adjacent nodes as adjacent edges. Extract the node features of each tower spatial diagram node.

[0024] S2 combines the tower space map and node characteristics, and generates a set of boundary constraint corridors through the strategic corridor identification method. Within each boundary constraint corridor, a comprehensive cost function is constructed.

[0025] In this embodiment, the step of generating a set of boundary-constrained corridors by combining the tower space map and node features and using the strategic corridor identification method specifically involves: Based on node characteristics, generate a recommended corridor width for each node; Based on node characteristics, the attention weights between nodes are calculated using a graph neural network model, and the strategic importance score of each node is output. By combining strategic importance scores with recommended corridor widths, a set of boundary constraint corridors is obtained.

[0026] In this embodiment, the recommended formula for calculating corridor width is as follows:

[0027] in, ,

[0028] In the formula, To recommend corridor width, To map the dimensionless score to the scale parameter of the actual physical length, For nodes The load-bearing capacity score For nodes Distance rating As the weight for the bearing capacity score, The weights for distance ratings satisfy the following conditions: , Minimum corridor width, Maximum corridor width, For amplitude limiting operators, For the geological bearing capacity of the node, For bearing capacity scaling parameters, The adjustment parameters are used to control the shape of the scoring function curve. For nodes Geographical distance to the known core area or center point of the plan. The parameter used to control the rate of exponential decay.

[0029] It should be noted that the geological bearing capacity of the nodes is obtained from geological survey and foundation bearing capacity test data, and the parameters controlling the exponential decay rate are calibrated according to the spatial distribution scale of typical work areas.

[0030] In this embodiment, the strategic importance scoring formula is as follows:

[0031] in,

[0032] In the formula, Score the strategic importance. For activation function, For the weights of the output layer, These are the bias parameters for the output layer. For nodes The set of neighboring nodes, neighboring nodes For nodes Attention weights For nodes eigenvectors, For nodes eigenvectors, For a trainable linear mapping matrix, To make nodes Feature vectors and nodes The feature vectors are concatenated to form joint features. This is the activation function.

[0033] The specific formula for the scoring loss function is as follows:

[0034] In the formula, The value of the scoring loss function. For the set of nodes in the tower space diagram, For nodes, Score the strategic importance. For nodes Training labels, The squared error function, The weights are the topological regularization terms. This is a topological regularization term.

[0035] It should be noted that the recommended corridor width refers to the lateral width of the corridor space that is recommended to be reserved near the node to ensure line safety, construction feasibility, and safe clearance when crossing obstacles.

[0036] In this embodiment, the step of combining strategic importance scores and recommended corridor widths to filter and obtain a set of boundary constraint corridors is specifically as follows: Nodes with strategic importance scores greater than a preset score threshold are selected to obtain a set of high-value nodes; Based on the recommended corridor width, the neighborhood is expanded with high-value nodes as the center to generate corresponding corridors; The corridors are merged and connected components are extracted to obtain a set of boundary-constrained corridors.

[0037] In this embodiment, nodes with strategic importance scores greater than a preset score threshold are selected to obtain a set of high-value nodes.

[0038] The high-value node set specifically includes:

[0039] In the formula, A collection of high-value nodes. This is a preset scoring threshold.

[0040] In this embodiment, the corridor calculation formula is as follows:

[0041] in,

[0042] In the formula, For corridor zone, It is a node in a set of high-value nodes. For nodes With nodes Spatial distance function between them For nodes The radius of the neighborhood centered on the center For nodes Recommended corridor width.

[0043] In this embodiment, the corridors are merged and connected components are extracted to obtain a set of boundary-constrained corridors, specifically: All corridor bands are joined together to form a preliminary corridor domain, and connected components are extracted from the preliminary corridor domain; Connected components are filtered and merged according to preset area and distance thresholds to obtain a set of boundary-constrained corridors. .

[0044] In this embodiment, constructing a comprehensive cost function within each boundary constraint corridor specifically involves: Extract nodal physical quantity data from each boundary constraint corridor, including tower physical quantities, pile foundation physical quantities, conductor physical quantities, and market data; Based on the physical quantity data of nodes, under physical constraints, formulas for tower construction cost, foundation construction cost, and line operation loss theory are established. The comprehensive cost function is obtained by summing the formulas for tower construction cost, foundation construction cost, and line operation loss theory.

[0045] In this embodiment, the physical constraints are specifically:

[0046] In the formula, For the vertical load transmitted from the base of the tower, For foundation bearing capacity, Based on the effective bearing area, This is for the safety factor.

[0047] In this embodiment, the formula for the tower construction cost is as follows:

[0048] In the formula, For the tower construction cost, For fixed construction costs, As a high cost coefficient, As an altitude index, Additional fee for tower type.

[0049] In this embodiment, the basic construction cost formula is specifically as follows:

[0050] In the formula, Basic construction costs, Basic fixed fee, This is the unit price for earthwork. For the excavation volume, For pile foundation price, This represents the number of pile foundations.

[0051] In this embodiment, the theoretical formula for line operation loss is as follows:

[0052] The theoretical formula for the actual length of the conductor is as follows:

[0053] The formula for the sag theory is:

[0054] In the formula, For cross-loss, For current crossing, For conductor resistivity, The cross-sectional area of ​​the conductor is... This is the actual length of the conductor. Annual operating hours, The unit price of electricity consumption, The length of the sag. The unit weight of the conductor. For horizontal span, For horizontal tension, For the first The lifecycle operating loss cost of each gear range.

[0055] In this embodiment, the comprehensive cost function is specifically formulated as follows:

[0056] In the formula, For the number of towers, For the tower construction cost, This is based on the construction cost.

[0057] S3, based on the boundary-constrained corridor, takes minimizing the comprehensive cost function as the optimization objective and obtains the target tower layout sequence through the coordinated iteration of Monte Carlo tree search and physical-guided neural network.

[0058] In this embodiment, the target tower layout sequence is obtained based on the boundary-constrained corridor, with the optimization objective of minimizing the comprehensive cost function. This is achieved through the collaborative iteration of Monte Carlo tree search and a physically guided neural network. Specifically: Based on the historical boundary constraint corridor, the physical guidance neural network is pre-trained; Within each boundary constraint corridor, a candidate sequence of pole placement is generated through Monte Carlo tree search, and the candidate sequence of pole placement is input into a pre-trained physical guidance neural network to obtain the predicted comprehensive cost; The candidate sequence of tower placement is used as the action space, and the predicted comprehensive cost is used as the state space. Under safety constraints, the predicted comprehensive cost is negatively mapped and used as a reward signal to be fed back to the Monte Carlo tree search process. The candidate sequence of tower placement is updated in the action space to obtain the target tower placement sequence.

[0059] In this embodiment, the physical guided neural network includes an input layer, a graph message passing layer, a differentiable physical equation module, and an output layer. The input layer receives node physical quantity data, and the graph message passing layer aggregates information based on the spatial topology of the towers. The differentiable physical equation module includes sag theory formulas, conductor actual length theory formulas, and line operation loss theory formulas. The output layer includes a linear activation function.

[0060] In this embodiment, the pre-training of the physical guidance neural network based on the historical boundary constraint corridor specifically involves: Based on the historical boundary constrained corridor, a physical guidance neural network is constructed using historical node physical quantity data and topological relationships as input; The residuals obtained from the differentiable physical equations are used as the physical loss function to train the physical guidance neural network and output the physical quantity data of the predicted nodes. Based on the predicted node physical quantity data, the predicted comprehensive cost is calculated using a comprehensive cost function.

[0061] In this embodiment, the physical loss function formula is as follows:

[0062] In the formula, The physical loss function value. The number of training samples, For physical residuals, For regularization terms, , This is the loss weight.

[0063] It should be noted that the physical residual term is obtained through automatic differentiation of the differentiable physical equation.

[0064] In this embodiment, under safety constraints, the step of negatively mapping the predicted comprehensive cost as a reward signal and feeding it back to the Monte Carlo tree search process to update the candidate tower layout sequence in the action space to obtain the target tower layout sequence is as follows: The reward signal is input as the return value into the Monte Carlo tree search, and the comprehensive score of the node is calculated using the UCT formula; Based on the cumulative updates of comprehensive scores and node access counts and action access counts, the node strategy of Monte Carlo tree search is continuously optimized; After iterative convergence, a target tower layout sequence that minimizes the predicted comprehensive cost is generated.

[0065] In this embodiment, based on Monte Carlo tree search, starting from the beginning of the search tree, child nodes are selected layer by layer according to the UCT formula until a node that has not been fully expanded is reached, thereby generating a candidate sequence for tower layout. The pre-trained physical guidance neural network is invoked to predict the comprehensive cost corresponding to the extended child nodes, and the predicted comprehensive cost value is obtained. The predicted comprehensive cost value is negatively mapped to obtain the reward signal, which is then used as the approximate return value for the child node. The reward signal is transmitted back layer by layer from the extended child node along the selected path to update the number of visits, the number of action visits, and the average reward value of each node on the path, so as to drive the iterative optimization of the search tree strategy and finally obtain the target tower layout sequence.

[0066] In this embodiment, the safety constraint formula is specifically as follows:

[0067] In the formula, For the height of the tower, For the first The ground height of the tower, To the tower The sag values ​​of adjacent spans, This refers to the minimum clearance required by design specifications.

[0068] In this embodiment, the UCT formula is specifically as follows:

[0069] In the formula, For comprehensive scoring, For action At the node Average return For nodes Number of visits, For action At the node Number of visits, To explore parameters.

[0070] Example 2, Figure 2 The present invention provides a power transmission and transformation project management and control system based on regional geological characteristics, comprising: The data modeling module is used to acquire regional power grid and geological data, construct tower spatial maps, and extract node features; The corridor constraint module is used to combine the tower space map and node characteristics, generate a set of boundary constraint corridors through the strategic corridor identification method, and construct a comprehensive cost function within each boundary constraint corridor; The iterative optimization module is used to obtain the target tower layout sequence based on the boundary-constrained corridor and with the goal of minimizing the comprehensive cost function. It obtains the target tower layout sequence through the coordinated iteration of Monte Carlo tree search and physical-guided neural network.

[0071] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0072] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0073] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0074] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0075] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0076] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A power transmission and transformation project management and control method based on regional geological characteristics, characterized in that, Includes the following steps: Acquire power grid and geological data for the target area, construct a spatial map of the towers, and extract node features; By combining the tower spatial map and node characteristics, a set of boundary constraint corridors is generated through the strategic corridor identification method, and a comprehensive cost function is constructed within each boundary constraint corridor. Based on the boundary-constrained corridor, with the goal of minimizing the comprehensive cost function, the target tower layout sequence is obtained through the coordinated iteration of Monte Carlo tree search and physical-guided neural network. The method of combining tower spatial maps and node features, and generating a set of boundary-constrained corridors through strategic corridor identification, specifically involves: Based on node characteristics, generate a recommended corridor width for each node; Based on node characteristics, the attention weights between nodes are calculated using a graph neural network model, and the strategic importance score of each node is output. By combining strategic importance scores and recommended corridor widths, a set of boundary-constrained corridors is obtained. The set of boundary-constrained corridors is obtained by combining strategic importance scores and recommended corridor widths, specifically as follows: Nodes with strategic importance scores greater than a preset score threshold are selected to obtain a set of high-value nodes; Based on the recommended corridor width, the neighborhood is expanded with high-value nodes as the center to generate corresponding corridors; The corridors are merged and connected components are extracted to obtain a set of boundary-constrained corridors; Within each boundary constraint corridor, a comprehensive cost function is constructed as follows: Extract nodal physical quantity data from each boundary constraint corridor, including tower physical quantities, pile foundation physical quantities, conductor physical quantities, and market data; Based on the physical quantity data of nodes, under physical constraints, formulas for tower construction cost, foundation construction cost, and line operation loss theory are established. The comprehensive cost function is obtained by summing the tower construction cost formula, the foundation construction cost formula, and the line operation loss theory formula. The boundary-constrained corridor, with minimizing the comprehensive cost function as the optimization objective, obtains the target tower layout sequence through the collaborative iteration of Monte Carlo tree search and physically guided neural network, specifically as follows: Based on the historical boundary constraint corridor, the physical guidance neural network is pre-trained; Within each boundary constraint corridor, a candidate sequence of pole placement is generated through Monte Carlo tree search, and the candidate sequence of pole placement is input into a pre-trained physical guidance neural network to obtain the predicted comprehensive cost; The candidate sequence of tower placement is used as the action space, and the predicted comprehensive cost is used as the state space. Under safety constraints, the predicted comprehensive cost is negatively mapped and used as a reward signal to be fed back to the Monte Carlo tree search process. The candidate sequence of tower placement is updated with action space to obtain the target tower placement sequence. The specific formula for the safety constraint is as follows: In the formula, For the height of the tower, For the first The ground height of the tower, To the tower The sag values ​​of adjacent spans, This refers to the minimum clearance required by design specifications. 2.The method according to claim 1, wherein, The physical guidance neural network includes an input layer, a graph message passing layer, a differentiable physical equation module, and an output layer. The input layer receives node physical quantity data, and the graph message passing layer aggregates information based on the spatial topology of the towers. The differentiable physical equation module includes the sag theory formula, the actual conductor length theory formula, and the line operation loss theory formula; the output layer includes a linear activation function. 3.The method according to claim 2, wherein, The pre-training of the physical guidance neural network based on the historical boundary constraint corridor is specifically as follows: Based on the historical boundary constrained corridor, a physical guidance neural network is constructed using historical node physical quantity data and topological relationships as input; The residuals obtained from the differentiable physical equations are used as the physical loss function to train the physical guidance neural network and output the physical quantity data of the predicted nodes. Based on the predicted node physical quantity data, the predicted comprehensive cost is calculated using a comprehensive cost function. 4.The method according to claim 3, wherein, Under safety constraints, the predicted comprehensive cost is negatively mapped and used as a reward signal to be fed back to the Monte Carlo tree search process. The candidate sequence of tower placement is updated with the action space to obtain the target tower placement sequence, specifically as follows: The reward signal is input as the return value into the Monte Carlo tree search, and the comprehensive score of the node is calculated using the UCT formula; Based on the cumulative updates of the comprehensive score and the number of node visits and action visits, the node strategy of Monte Carlo tree search is continuously optimized; After iterative convergence, a target tower layout sequence that minimizes the predicted comprehensive cost is generated.

5. A system for managing and controlling power transmission and transformation projects based on regional geological characteristics as described in any one of claims 1-4, comprising: The data modeling module is used to acquire regional power grid and geological data, construct tower spatial maps, and extract node features; The corridor constraint module is used to combine the tower space map and node characteristics, generate a set of boundary constraint corridors through the strategic corridor identification method, and construct a comprehensive cost function within each boundary constraint corridor; The iterative optimization module is used to obtain the target tower layout sequence based on the boundary-constrained corridor and with the goal of minimizing the comprehensive cost function. It obtains the target tower layout sequence through the coordinated iteration of Monte Carlo tree search and physical-guided neural network.

Images