A power transmission line scheme generation and optimization method and system based on multi-agent cooperation

Abstract

The present application belongs to the technical field of electric power engineering design, and particularly relates to a power transmission line scheme generation and optimization method and system based on multi-agent collaboration. The method comprises: constructing a collaborative system composed of a regulation understanding agent, a space exploration agent and a structure generation agent. The regulation understanding agent analyzes the specifications and geographic data, outputs enhanced geographic space data with semantic labels, and generates a dynamic compliance energy field. The space exploration agent conducts path exploration under the real-time guidance of the energy field, and its fitness function integrates performance indicators and compliance energy path integrals. The structure generation agent generates a tower structure scheme based on the path, load and regulation constraints. After the scheme is verified by simulation, the performance data is fed back to the agent to drive closed-loop optimization. The present application realizes deep real-time collaboration of compliance judgment and optimization search, and improves design efficiency and scheme quality.

Description

Technical Field

[0001] This invention belongs to the field of power engineering design technology, specifically relating to a method and system for generating and optimizing transmission line schemes based on multi-agent collaboration. Background Technology

[0002] With the development of digitalization in power transmission line design, methods for automatic route planning using multi-source geographic data and optimization algorithms have emerged. For example, patent document CN119540443A discloses a typical scheme: it first acquires multi-source data such as laser point clouds and oblique photography to generate a 3D geographic model of the region; then, based on optimization methods such as genetic algorithms, it generates an initial route between the starting and ending points, primarily with economic efficiency as the objective; the scheme then performs segmented evaluation and manual modification of the initial route, and conducts tower placement and safety verification, finally outputting an optimized scheme. This method represents a widely adopted technical implementation path in the industry, based on multi-source geographic information data, generating a preliminary scheme through optimization algorithms, and then performing post-implementation compliance and safety verification.

[0003] However, the above scheme has a significant drawback: its optimization search process and the in-depth compliance judgment of regulations are disconnected. Specifically, in the core iterative search process, the optimization algorithm only uses explicit indicators such as length and cost as optimization targets, and cannot obtain in-depth compliance feedback from the regulatory analysis model in real time with each path iteration adjustment. The optimization process and compliance judgment are separated into two independent stages. This serial processing mode results in the optimization algorithm's search direction lacking real-time guidance from the regulatory level. A large amount of computational resources are consumed in exploring path regions that will be judged as non-compliant in the future, resulting in low search efficiency and difficulty in automatically and efficiently converging to a globally compliant and economically optimal solution under complex multi-regulation constraints. Summary of the Invention

[0004] The technical problem to be solved by this invention is to overcome the serious disconnect between artificial intelligence and the optimization search process.

[0005] To address the aforementioned technical problems, the first aspect of this invention discloses a method for generating and optimizing transmission line schemes based on multi-agent collaboration, the method comprising: Construct a multi-agent collaborative system comprising a specification understanding agent, a space exploration agent, and a structure generation agent; The specification understanding agent parses and models the specification text to form a specification knowledge network, and interprets multi-source geographic information data based on this network to output enhanced geospatial data with semantic tags. The space exploration agent performs parallel exploration within the constrained space defined by the enhanced geospatial data, generating candidate path schemes in real time. In each exploration and evaluation, the specification understanding agent is synchronously invoked to perform compliance deduction and risk quantification on the path segments, and the results are used as feedback signals to dynamically adjust the search strategy of the space exploration agent; The structure generates an intelligent agent that, based on the optimized path scheme and combined with terrain and load data, automatically generates a three-dimensional structural scheme for the tower and foundation under the design constraints output by the specification understanding agent. The generated structural scheme is simulated and verified, and the simulation performance data is fed back to the specification understanding agent to drive the multi-agent collaborative system to further optimize the scheme that does not meet the performance threshold.

[0006] As an optional implementation, in the first aspect of the present invention, the process by which the specification understanding agent performs compliance deduction and risk quantification on path segments specifically involves generating a dynamic compliance energy field. For a point on the path Compliance energy value The calculation formula is: ; in, Point Violation of Article The degree of constraint of class specifications, For the preset nonlinear mapping function, These are dynamic weighting coefficients that are relevant to the current project context.

[0007] As an optional implementation, in the first aspect of the present invention, the space exploration agent employs a multi-objective optimization algorithm, and its fitness function F is constructed as follows: ; in, For path The comprehensive performance index function, For compliant energy fields along the path The path integral is minimized to deeply embed compliance constraints into the optimization objective.

[0008] As an optional implementation, in the first aspect of the present invention, before activating the space exploration agent, the specification understanding agent pre-generates a high-compliance guidance channel in the enhanced geospatial data based on core mandatory constraints, as initialization prior information for the space exploration agent.

[0009] As an optional implementation, in the first aspect of the present invention, the structure generating agent is constructed based on a conditional generation model. Its generation process takes path profiles, load conditions, and design constraint latent vectors extracted from the specification understanding agent as conditional inputs, and outputs parameterized structural schemes.

[0010] As an optional implementation, in the first aspect of the present invention, after generating the structural scheme, a finite element analysis engine is called to perform mechanical verification. If the verification fails, the stress exceeding the standard information and the original design constraints are fed back to the structure generation agent to drive iterative modification of the local structure.

[0011] As an optional implementation, in the first aspect of the present invention, when the simulation performance data fails to meet the requirements, it is converted into a structured performance specification fragment and added to the input knowledge network of the specification understanding agent to guide the generation of subsequent schemes.

[0012] As an optional implementation, in the first aspect of the present invention, it further includes: deploying a lightweight version of the specification understanding agent and the space exploration agent on a mobile terminal, combining augmented reality devices to visualize and interactively adjust the on-site solution, and adjusting the data feedback to update the central system model.

[0013] A second aspect of this invention discloses a transmission line scheme generation and optimization system based on multi-agent cooperation, the system comprising: The system construction module is used to build a multi-agent collaborative system that includes a specification understanding agent, a space exploration agent, and a structure generation agent; The specification understanding module is used to parse and model the specification text through the specification understanding agent to form a specification knowledge network, and interpret multi-source geographic information data based on this network to output enhanced geospatial data with semantic tags. The path exploration module is used to perform parallel exploration within the constrained space defined by the enhanced geospatial data through the spatial exploration agent, and generate candidate path schemes in real time. The compliance deduction module is used to synchronously call the specification understanding agent to perform compliance deduction and risk quantification on the path segments in each exploration assessment, and use the results as feedback signals to dynamically adjust the search strategy of the space exploration agent; The structure generation module is used to generate a three-dimensional structural scheme for the tower and foundation based on the optimized path scheme and combined with terrain and load data, driven by the design constraints output by the specification understanding agent. The simulation verification module is used to perform simulation verification on the generated structural scheme, feed back the simulation performance data to the specification understanding agent, and drive the multi-agent collaborative system to further optimize the scheme that does not meet the performance threshold.

[0014] A third aspect of this invention discloses another transmission line scheme generation and optimization system based on multi-agent cooperation, the system comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the method for generating and optimizing transmission line schemes based on multi-agent cooperation disclosed in the first aspect of the present invention.

[0015] The fourth aspect of the present invention discloses a computer-readable storage medium storing computer instructions, which, when invoked by a processor, are used to execute a method for generating and optimizing transmission line schemes based on multi-agent cooperation disclosed in the first aspect of the present invention.

[0016] Compared with the prior art, the beneficial effects of the present invention are: By constructing a multi-agent collaborative system, real-time interaction and deep collaboration between optimization search and compliance judgment are achieved during path optimization. In each path iteration, this method synchronously invokes a specification understanding agent to perform compliance deduction and risk quantification for the current path segment, and uses the quantified compliance cost as immediate feedback to dynamically adjust the search direction. This enables the optimization process to proactively avoid non-compliant areas, reduce exploration in the invalid solution space, and improve search efficiency. Thus, under complex multi-specification constraints, it more efficiently finds a global optimization solution that balances compliance, economy, and feasibility. Simultaneously, by automatically feeding back simulation verification results and driving targeted re-optimization of substandard solutions, a complete closed loop from design, verification, to correction is formed. Attached Figure Description

[0017] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating a method for generating and optimizing transmission line schemes based on multi-agent collaboration, as disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of a transmission line scheme generation and optimization system based on multi-agent collaboration disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of another transmission line scheme generation and optimization system based on multi-agent collaboration disclosed in an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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.

[0019] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0020] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0021] This invention discloses a method and system for generating and optimizing transmission line schemes based on multi-agent collaboration. By constructing a multi-agent collaborative system, real-time interaction and deep collaboration between optimization search and compliance judgment are achieved during the path optimization process. In each path iteration exploration, the method synchronously invokes a specification understanding agent to perform compliance deduction and risk quantification for the current path segment, and uses the quantified compliance cost as immediate feedback to dynamically adjust the search direction. This enables the optimization process to proactively avoid non-compliant areas, reduce exploration in the invalid solution space, and improve search efficiency. Thus, under complex multi-specification constraints, it more efficiently finds a global optimization scheme that balances compliance, economy, and feasibility. Simultaneously, by automatically feeding back simulation verification results and driving targeted re-optimization of non-compliant schemes, a complete closed loop from design, verification, to correction is formed.

[0022] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for generating and optimizing transmission line schemes based on multi-agent collaboration, as disclosed in an embodiment of the present invention. Figure 1The described method for generating and optimizing transmission line schemes based on multi-agent collaboration is applied to data processing chips, processing terminals, or processing servers. The processing server can be a local server or a cloud server; this embodiment of the invention does not impose any limitations. Figure 1 As shown, the method for generating and optimizing transmission line schemes based on multi-agent collaboration can include the following operations: 101. Construct a multi-agent collaborative system that includes a specification understanding agent, a space exploration agent, and a structure generation agent.

[0023] Specifically, this multi-agent collaborative system achieves coordination and optimization of the task execution process by resolving task constraints through specification understanding agents, searching feasible solution domains through spatial exploration agents, and synthesizing output results through structure generation agents.

[0024] It is evident that the system achieves its goal of enhancing task processing capabilities and improving output quality through the division of labor and cooperation among multiple agents.

[0025] 102. The specification understanding agent parses and models the specification text to form a specification knowledge network, and interprets multi-source geographic information data based on this network to output enhanced geospatial data with semantic tags.

[0026] Specifically, by establishing a specification knowledge network, a structured interpretation basis is provided for multi-source geographic information data, enabling the data to be automatically assigned explicit semantic information, thereby forming enhanced geospatial data.

[0027] As can be seen, this invention achieves the effect of systematically generating geographic information results rich in semantic knowledge from unstructured text and raw data.

[0028] 103. The space exploration agent performs parallel exploration within the constrained space defined by the enhanced geospatial data and generates candidate path schemes in real time.

[0029] Specifically, this step enables the spatial exploration agent to conduct parallel exploration within the semantic constraint space, thereby achieving real-time generation of candidate path solutions and improving the efficiency and diversity of path discovery.

[0030] As can be seen, this invention achieves the effect of rapidly generating multiple feasible path alternatives under clear geographical semantic constraints.

[0031] 104. In each exploration evaluation, the specification understanding agent is synchronously invoked to perform compliance deduction and risk quantification of the path segment, and the results are used as feedback signals to dynamically adjust the search strategy of the space exploration agent.

[0032] Specifically, by conducting real-time compliance simulations and risk quantifications, and using the results as feedback to dynamically adjust search strategies, the path exploration process can promptly avoid risks and comply with regulatory constraints.

[0033] As can be seen, the present invention achieves the effect of simultaneously completing compliance optimization during the path generation process and guiding the search direction to converge toward a more feasible region.

[0034] 105. Using the structure to generate an intelligent agent, based on the optimized path scheme and combined with terrain and load data, and driven by the design constraints output by the specification understanding intelligent agent, automatically generate a three-dimensional structural scheme for the tower and foundation.

[0035] Specifically, based on the route plan, terrain and load data, and driven by design constraints, the automatic generation of three-dimensional structural schemes for towers and foundations was achieved, improving the automation level of structural design.

[0036] As can be seen, this invention achieves the effect of combining route path, environmental data and design specifications to automatically output a matching three-dimensional structural design scheme.

[0037] 106. The generated structural scheme is simulated and verified, and the simulation performance data is fed back to the specification understanding agent to drive the multi-agent collaborative system to further optimize the scheme that does not meet the performance threshold.

[0038] Specifically, performance data is generated through simulation verification and fed back to the specification understanding agent, driving the system to further optimize solutions that do not meet the threshold, forming a closed-loop iteration based on performance verification.

[0039] As can be seen, the present invention achieves the effect of automatically driving the scheme iteration through simulation feedback data, thereby systematically improving the performance and reliability of the final output scheme.

[0040] As an optional embodiment, the process by which the specification understanding agent performs compliance deduction and risk quantification on path segments in the above steps specifically involves generating a dynamic compliance energy field. For a point on the path Compliance energy value The calculation formula is: ; in, Point Violation of Article The degree of constraint of class specifications, For the preset nonlinear mapping function, These are dynamic weighting coefficients that are relevant to the current project context.

[0041] In this embodiment of the invention, the traditional yes / no compliance judgment is transformed into a continuous and quantifiable spatial energy distribution, thereby providing real-time and precise navigation signals for the optimization algorithm. This mechanism solves the problem of blindness caused by the lack of gradient guidance during the optimization search process. Specifically, by calculating the compliance energy value of each point on the path, the algorithm can not only identify whether a violation has occurred, but also perceive the severity and nature of the violation. Therefore, during iteration, it automatically tends to select path directions with lower energy values, i.e., better compliance, thus improving search efficiency and the quality of the final solution.

[0042] This invention defines a core computational model, namely the compliance energy field. This model is used to perform real-time, quantitative compliance assessment of any path solutions generated during the optimization process; The compliance energy field is a virtual field covering the entire geographical space. Every point within the field... Each has a corresponding scalar value This value is called the compliance energy value at that point. The level of this value directly reflects the level of compliance risk associated with constructing the line at that point; a higher energy value indicates a greater risk of violating relevant regulations and standards. point Compliance energy value It is not the result of a single judgment, but rather a result derived through a comprehensive and weighted calculation of violations of all relevant regulatory constraints. The calculation formula is as follows: ; degree of violation This is a basic input representing points on the path. For the The degree of violation of a class-specific constraint. It is a quantifiable value; for example, when the constraint is "at least 100 meters from residential areas," It can be the difference between the actual distance and 100 meters (either in absolute value or after normalization). The larger the value, the more severe the violation. Nonlinear mapping function This is a predefined mathematical function used to determine the degree of violation of the original. The conversion is performed. It has two functions: first, it maps the degree of violation with different physical meanings (such as distance, elevation difference, angle) to a unified and comparable dimension; second, it can introduce nonlinear characteristics. For example, when the degree of violation exceeds a certain critical point, its energy contribution can be amplified by a function to reflect the "hardness" or sensitivity of certain constraints. Dynamic weighting coefficients This is a key parameter reflecting the agent's ability to "understand" and "be aware of context." Weighting coefficients. It is not fixed, but dynamically generated by the specification understanding agent based on the specific context of the current project. The context may include factors such as the project's location, voltage level, and priority protection targets. For example, in a project in an ecologically sensitive area, constraints related to environmental protection... Will get higher The value makes this type of violation appear in the total energy. The proportion of the middle is larger, thus strongly guiding the path to avoid this area; Weighted summation Ultimately, all of them The contribution value of class constraints after the above processing ( Summing these values ​​yields the total compliance energy value for that point. This value comprehensively reflects the point. Overall compliance risk under all constraints.

[0043] It is evident that through this mechanism, complex rules and regulations are transformed into an energy field that can be calculated and differentiated by machines in real time, enabling optimization algorithms to search with minimizing the total compliance energy integral of the path as one of their explicit objectives. This achieves a deep and dynamic integration of normative knowledge into the automated design process.

[0044] As an optional embodiment, in the above steps, the space exploration agent employs a multi-objective optimization algorithm, and its fitness function F is constructed as follows: ; in, For path The comprehensive performance index function, For compliant energy fields along the path The path integral is minimized to deeply embed compliance constraints into the optimization objective.

[0045] In this embodiment of the invention, the compliance energy field is directly embedded into one of the core objectives of the optimization algorithm through mathematical form, realizing multi-objective deep collaborative optimization of compliance constraints and path comprehensive performance. This method overcomes the limitations of traditional optimization where compliance is treated as an external hard boundary or a post-validation term, enabling the optimization process to automatically weigh the economic efficiency, engineering performance, and compliance risks of the path in each iteration, thereby guiding the search process to converge directly to a Pareto front solution that achieves both excellent comprehensive performance and compliance.

[0046] This invention clarifies the specific construction method of the optimization objective function of a space exploration agent. Its core is to quantify the aforementioned compliant energy field into a computable cost term and combine it with traditional performance indicators to form a unified objective function to drive the search.

[0047] fitness function Constructed as the difference between two terms: This structure is designed to maximize fitness. Value. To achieve this goal, the optimization process needs to simultaneously pursue maximizing path performance. Compliance energy accumulated with the minimum path .

[0048] Detailed explanation of the formula components: Path performance function This item represents the path. The comprehensive evaluation of traditional engineering optimization objectives is usually a positive indicator that needs to be maximized. Its specific form can be defined according to project requirements, and commonly includes a weighted combination of sub-objectives such as total line length, total cost estimate, and construction difficulty coefficient. The higher the value, the better the path is in terms of pure performance metrics.

[0049] Compliant Energy Path Integral This is the key to the deep embedding and optimization of compliance in this invention. It represents the compliance energy field. Along the entire path The line integral (or path integral) is the sum of the compliance energy values ​​of all points traversed along the path. Physically, this integral value represents the overall compliance risk level of the entire path; a higher integral value indicates a deeper level of non-compliance or higher risk. In the fitness function, this term appears as a decrease, signifying a higher cumulative compliance risk on the path and a greater impact on fitness. The greater the negative impact, the better.

[0050] The significance of integral operations: integral operations This achieves a convergence from "compliance energy at a point" to "compliance cost along a line." It ensures that the optimization algorithm focuses not only on the compliance of individual isolated points but also on the global compliance performance of the entire path. The optimization algorithm attempts to shorten the path length (potentially improving...). If this causes the path to traverse a high compliance energy region (leading to a dramatic increase in the integral term), then the overall fitness will decrease. This could actually decrease the path, thus preventing the search from proceeding in that direction. This forces the algorithm to find a compromise path that balances length with the risk of detours. As can be seen, ultimately, the fitness function As a single scalar, it unifies the two dimensions of performance and compliance. This is achieved through algorithm design to maximize... To search for targets, the space exploration agent does not need to make complex external trade-offs among multiple independent targets. Instead, in each path mutation and selection, the function automatically performs an instant calculation and comparison of the "performance improvement benefits" and the "increased costs of compliance risks." This allows compliance knowledge to be transmitted through the energy field. The integral term is transformed into a search gradient that can be naturally understood and utilized by the optimization algorithm, achieving true deep collaboration and real-time guidance.

[0051] As an optional embodiment, in the above steps, before activating the space exploration agent, the specification understanding agent pre-generates a high-compliance guidance channel in the enhanced geospatial data based on core mandatory constraints, as the initialization prior information of the space exploration agent.

[0052] In this embodiment of the invention, the pre-generated guiding channel mechanism provides a high-quality initial solution and a structured search space for subsequent intelligent optimization searches, thereby significantly improving the convergence speed and solution quality of the optimization process. This method, through pre-emptive intelligent reduction analysis, transforms the most important mandatory compliance knowledge into prior structured information about the search space, effectively avoiding the inefficiency caused by completely random or blind searches in the initial stages of the optimization algorithm. It ensures that the search starts from a high-potential region, accelerating the approach to the global optimum. Specifically, it includes: 1. Purpose and Location of this Step: This step occurs before the main iteration loop of the spatial exploration agent (such as an evolutionary algorithm or reinforcement learning agent). Its purpose is to use known rule knowledge to perform preliminary and intelligent screening and delineation of the large and complex solution space (i.e., the set of all possible paths), narrowing the search range and setting a favorable starting point for the optimization algorithm.

[0053] 2. Identification of Core Mandatory Constraints: First, the specification understanding agent analyzes the constructed specification knowledge network to identify "core mandatory constraints." These constraints typically refer to clauses that, if violated, will directly lead to the rejection of the solution and leave no room for compromise, such as prohibitions on traversing core nature reserves, military restricted areas, or major adverse geological formations. The agent can understand the precise geographical boundaries and legal strength of these constraints.

[0054] 3. Generation of High-Compliance Guiding Channels: Based on identified core mandatory constraints, the specification understanding AI performs spatial analysis on augmented geospatial data. Its core logic is to find connected spaces between the starting and ending points that completely avoid all core prohibited areas. This process does not generate a precise path, but rather one or more "channel" regions. These "guiding channels" can be understood as relatively safe strip-shaped or regional spaces outlined on a 2D or 3D map, ensuring that path searches within these channels will at least avoid the most critical compliance red lines. The generation methods may involve geocomputing techniques such as spatial buffer analysis, obstacle expansion, and connected domain calculation.

[0055] 4. Application as initialization prior information: The generated high-compliance guidance channel is passed to the space exploration agent as its "initialization prior information." Specific applications include, but are not limited to: Initial population generation: If the space exploration agent uses a population optimization algorithm (such as a genetic algorithm), the initial population of the algorithm (i.e. the first batch of candidate paths) will be restricted to random or rule-based generation within the guidance channel to ensure that all initial solutions satisfy the core mandatory constraints.

[0056] Search space constraints: The mutation, crossover, and other operations of the optimization algorithm can be constrained so that the candidate solutions generated mainly evolve around this channel, thereby greatly reducing invalid searches.

[0057] Heuristic information: For agents based on sampling or reinforcement learning, this channel can serve as a basis for initializing the sampling distribution or value function, guiding the agent to prioritize exploring high-value regions. It is evident that through this preliminary step, the symbolic and logical knowledge of the specification understanding agent is transformed and injected into the data-driven and randomized search process of the spatial exploration agent, achieving an efficient connection between knowledge-driven and data-driven approaches, and ensuring the efficiency of the optimization process from the very beginning.

[0058] As an optional embodiment, in the above steps, the structure generating agent is constructed based on a conditional generation model. Its generation process takes path profiles, load conditions, and design constraint latent vectors extracted from the specification understanding agent as conditional inputs, and outputs parameterized structural schemes.

[0059] In this embodiment of the invention, by introducing a conditional generation model and using multi-source information as input, this method overcomes the inefficiencies and lack of flexibility caused by relying on standard libraries and manual selection in traditional design. It can instantly generate customized structural schemes that are compatible with mechanical, electrical, and regulatory standards for each unique optimization path, achieving deep integration and collaboration between the line path and the tower structure during the design phase.

[0060] This invention describes the working principle of an intelligent structural scheme automatic generator. Instead of selecting models from a fixed library, it dynamically generates new design schemes based on specific scenarios. Specifically, it includes: 1. Core Generation Mechanism: The core of the structure-generating agent is a conditional generation model. This is an artificial intelligence model characterized by its ability to generate new data samples that meet given conditional information. In this application, instead of generating images or text, it generates parameterized engineering structure schemes.

[0061] 2. Detailed Explanation of Conditional Inputs: The accuracy and adaptability of the generation process depend on the following three dimensions of conditional inputs, which together define all the requirements that the generated structure must meet: Path profile: This is the spatial trajectory information of the guide line suspension point, usually including a series of three-dimensional coordinate points. It determines the longitudinal tension, rotation angle, and height difference between adjacent towers that the tower needs to withstand, and is the most direct factor determining the structural stress and height.

[0062] Load conditions: These include the various external forces that the tower structure must withstand due to the environmental requirements of the line project, such as wind load, ice load, seismic action, and the weight and tension of the conductor itself. These conditions are the direct basis for structural strength and stability design.

[0063] Design constraint implicit vector: This is the core knowledge transmitted from the specification understanding agent. After analyzing the relevant design specifications, the specification understanding agent transforms the textual clauses into a machine-readable, dense numerical vector. This vector encodes rich structured design rules and constraints regarding material grades, construction requirements, safety factors, dimensional limits, and more. It ensures that the generated structural scheme fundamentally conforms to the specification provisions, and not merely achieves mechanical feasibility.

[0064] 3. Output and Workflow: After receiving the above conditions, the model outputs a parameterized structural scheme through its complex internal neural network mapping. This output is not an uneditable 3D mesh file, but a set of key parameters defining the structural type, main dimensions, material specifications, and component arrangement. For example, for a tower, the output might include parameterized descriptions such as tower type, nominal height, root opening size, and main material specifications. These parameters can directly drive 3D modeling software to instantiate a specific model or enter subsequent automated analysis processes. It is evident that this working method enables context-based design, ensuring that each tower is tailored to its specific location, thereby improving the economy and compliance of the design from the source. This represents a key shift from traditional standardized design to intelligent customized design.

[0065] As an optional embodiment, in the above steps, after generating the structural scheme, the finite element analysis engine is called to perform mechanical verification. If the verification fails, the stress exceeding the standard information and the original design constraints are fed back to the structure generation agent to drive iterative modification of the local structure.

[0066] In this embodiment of the invention, the separate generation, analysis, and modification stages in traditional design are integrated into a self-driven optimization process, thereby improving the reliability and efficiency of structural design. This mechanism ensures that each generated structural scheme can pass rigorous mechanical verification and accurately strengthens substandard parts through intelligent feedback, automatically achieving safety goals without human intervention, and realizing full automation from concept generation to qualified finished product.

[0067] This invention describes an embedded, automated verification and optimization sub-process after the initial scheme of structural intelligent agent generation. This process aims to ensure that the generated structure is not only compliant with rules but also absolutely safe in terms of physical performance, specifically including: 1. Initiation of Automated Verification: After the structure generation agent outputs a parameterized structural scheme, the system does not treat this scheme as the final result, but automatically triggers subsequent verification steps. Based on the parameters of the scheme, the system automatically calls the finite element analysis engine in the background to generate the corresponding computational model. This model can accurately simulate the mechanical response of the tower under various load conditions.

[0068] 2. Core Criteria for Mechanical Verification: The finite element analysis engine performs static and dynamic calculations on the structure and outputs results such as stress, strain, and displacement of key components. The core criteria for passing the verification are usually whether the maximum equivalent stress exceeds the allowable stress of the material or whether the deformation exceeds the allowable displacement. These criteria are derived from design specifications and are pre-configured in numerical form. For example, verification conditions can be quantified as: [The following is a list of requirements for compliance with mechanical verification standards:] ,in The maximum equivalent stress obtained from finite element analysis. This represents the allowable stress of the material.

[0069] 4. Formation of Intelligent Feedback: If the verification fails, the system will perform intelligent diagnosis instead of simply reporting an error. It will locate the specific component where the stress exceeds the limit and extract the specific value of the excess (e.g., the stress at a certain point exceeds the allowable value by 15%). This "stress exceeding the limit information" is encapsulated together with the implicit design constraint vectors initially generated by the driver into a new, enhanced feedback condition package.

[0070] 5. Iterative Modification Mechanism Driven by the System: This feedback condition package is resent to the structural generative agent, triggering a new generation iteration. In this iteration, the agent's generation conditions change: it must not only satisfy the original path, load, and specification constraints, but also additionally satisfy the new instruction to "repair stress overshoot at a specified location." The generative model understands this feedback and, while maintaining other conditions, enhances the local component by adjusting its size, shape, or material parameters, thereby generating an improved version of the solution. This process is repeated until finite element analysis confirms that all mechanical properties meet the requirements.

[0071] It is evident that this closed loop transforms structural design and verification from a traditional, serial, and manual process into a parallel, automated, and intelligent optimization process, ensuring that the final output solution is both innovative and engineering reliable.

[0072] As an optional embodiment, in the above steps, when the simulation performance data does not meet the requirements, it is converted into a structured performance specification fragment and added to the input knowledge network of the specification understanding agent to guide the generation of subsequent schemes.

[0073] In this embodiment of the invention, a closed-loop learning path is established from physical performance verification results to a design knowledge system. This method transcends the unidirectional process of simulation used only for verification in traditional design, enabling the experience of substandard performance to be extracted and precipitated into structured new knowledge by the system. This allows the agent to automatically avoid similar problems in subsequent design iterations, realizing the system's self-evolution and continuous accumulation of design experience, fundamentally improving the long-term adaptability of the design system and the first-time success rate of the solution.

[0074] This invention discloses a key mechanism that enables the entire system to possess continuous learning capabilities. It transforms one-time simulation verification results specific to a particular scheme into reusable, general knowledge that can guide the generation of all future schemes. Specifically, this includes: 1. Triggering and Input of the Mechanism: This mechanism is triggered when the output data of the digital twin simulation (such as electromagnetic simulation, dynamic response analysis) indicates that certain performance indicators of the current scheme (such as electric field strength, wind vibration response amplitude) do not meet the preset safety threshold. Its original input is the original simulation data, such as the tower number that does not meet the requirements, the specific amount exceeding the standard, the location of the exceeding standard, etc.

[0075] 2. Structured Transformation Process: This is the core step of the mechanism. The system does not simply record a conclusion such as "the electrical clearance of tower N in a certain scheme is insufficient," but performs an intelligent transformation. It abstracts, summarizes, and reformulates the specific performance defect into a machine-parseable, specification-like structured knowledge unit, namely, a "performance specification fragment."

[0076] Structured Form: This fragment may be represented in the form of frames, triples, or rules with attributes. For example, a specific performance defect might be transformed into a structured rule: "In the context of [terrain features: near the slope] and [meteorological conditions: strong wind area], for [tower type: cat-head tower], its [design parameter: crossarm length] should satisfy [constraint: less than]..." The new requirement is to ensure that the performance indicator: electrical clearance after wind deflection meets the standard. Among other things, It is an empirical critical value calculated by reverse engineering based on the data of this exceedance.

[0077] 3. Integration of Knowledge Networks: The generated structured performance specification fragments are added as new knowledge nodes to the dynamically scalable knowledge network maintained by the specification understanding agent. This network originally stores clauses parsed from the text specification. The newly added fragments establish semantic associations with existing related nodes in the network (such as text specifications about "electrical clearance" and classifications about "terrain"). This integrates performance knowledge with text specification knowledge, jointly forming the knowledge foundation for the agent's reasoning and decision-making.

[0078] 4. Guiding Principles for Subsequent Generation: When generating structural solutions for new paths or similar environments, the specification understanding agent will retrieve all relevant constraints from the expanded knowledge network, including newly added performance specification fragments. When the agent recognizes that the current design scenario (such as "bevel" or "strong wind") matches the applicable context of a certain performance fragment, the empirical constraints implied in that fragment (such as "the crossarm length must be shorter than...") will be applied. This condition will be activated and passed to the structure-generating agent as a strong constraint. Thus, the new generation of design solutions inherits the experience gained from the trial-and-error process of the previous generation, proactively fulfilling requirements that would otherwise only be discovered through later simulations, thereby avoiding the recurrence of similar problems. It is evident that the system transforms discrete, passive simulation problem feedback into systematic, proactive prior design knowledge, thereby achieving continuous growth in design intelligence.

[0079] As an optional embodiment, the above steps also include: deploying a lightweight version of the specification understanding agent and the space exploration agent on a mobile terminal, combining augmented reality devices to visualize and interactively adjust the on-site solution, and adjusting the data feedback to update the central system model.

[0080] In this embodiment of the invention, the intelligent design system is extended from the backend server to the engineering site, bridging the "last mile" between digital design and physical surveying and verification. This method solves the problems of disconnect between the design studio's plans and the complex realities of the site, as well as the lag in adjustment feedback, in the traditional process. Through real-time visualization and natural interaction, the experience and judgment of on-site experts can be integrated into the optimization process in real time, significantly improving the feasibility and decision-making efficiency of the design plan, and continuously enhancing the adaptability of the central model using real-world data.

[0081] This invention describes an application architecture that supports on-site collaborative design and continuous system learning. It extends core intelligent capabilities to field operating environments and establishes a feedback loop from the field back to the design center, specifically including: 1. Lightweight Deployment: To run on mobile terminals with limited computing power, storage, and power consumption, the model of the specification understanding agent and the space exploration agent needs to be lightweighted. This is not a simple reduction of functionality, but rather a significant reduction in model size and computational overhead through techniques such as knowledge distillation, model pruning, and quantization, while maintaining core reasoning and decision-making capabilities as much as possible. This results in a lightweight inference engine that can be embedded in mobile applications or dedicated devices. After deployment, the field terminal has the basic capability to perform real-time intelligent analysis in offline or weak network environments.

[0082] 2. Augmented Reality Visualization: Mobile terminals and augmented reality devices work collaboratively. The terminal uses its camera to capture real-time video streams of the surrounding environment. Simultaneously, a lightweight intelligent agent, based on the real-time positioning and attitude information obtained by the terminal, aligns and 3D registers pre-issued or instantaneously generated solution data (such as recommended paths, tower locations, and risk heat maps), and precisely overlays and renders it onto the real video footage in the form of virtual 3D models, lines, or colored heat maps. This allows surveyors to intuitively see the effect of the design scheme "landing" on the real terrain from a first-person perspective, and to readily assess the rationality of the scheme.

[0083] 3. Interactive Adjustment: When on-site personnel, based on AR visualization and the actual situation, deem the plan needs adjustment, they can modify it through intuitive and natural interactive methods. For example, they can change the direction by dragging virtual path points in the AR screen with gestures, or mark "undetected obstacles exist here" using voice commands. The interactive commands are captured by the terminal and converted into structured adjustment data.

[0084] 4. Data Feedback and Model Updates: Interactive adjustment data generated on-site, along with supplementary environmental data collected by terminals in the field, are synchronously fed back to the central server. This data is extremely valuable, as it records constraints encountered by the design scheme in the real world that were not considered in the original model or were missing data. The central system uses this feedback data to incrementally train or fine-tune all specification understanding agents and space exploration agents, enabling the next design scheme generation to automatically avoid similar problems or better reflect the actual conditions of a specific type of site. This constitutes a complete enhanced closed loop from virtual to physical and then from physical feedback back to virtual.

[0085] It is evident that this mechanism not only improves the efficiency and quality of a single survey and design, but also enables the entire intelligent system to continuously evolve by absorbing real-world feedback from the field, becoming increasingly compatible with complex engineering realities.

[0086] Example 2 Please see Figure 2 , Figure 2This is a schematic diagram of a transmission line scheme generation and optimization system based on multi-agent collaboration disclosed in an embodiment of the present invention. Figure 2 The described multi-agent collaborative transmission line scheme generation and optimization system can be applied to data processing chips, processing terminals, or processing servers. The processing server can be a local server or a cloud server; this embodiment of the invention does not impose any limitations. Figure 2 As shown, the transmission line scheme generation and optimization system based on multi-agent collaboration may include the following operations: System construction module 201 is used to build a multi-agent collaborative system that includes a specification understanding agent, a space exploration agent, and a structure generation agent.

[0087] Specifically, this multi-agent collaborative system achieves coordination and optimization of the task execution process by resolving task constraints through specification understanding agents, searching feasible solution domains through spatial exploration agents, and synthesizing output results through structure generation agents.

[0088] It is evident that the system achieves its goal of enhancing task processing capabilities and improving output quality through the division of labor and cooperation among multiple agents.

[0089] The specification understanding module 202 is used to parse and model the specification text through the specification understanding agent to form a specification knowledge network, and interpret multi-source geographic information data based on this network to output enhanced geospatial data with semantic tags.

[0090] Specifically, by establishing a specification knowledge network, a structured interpretation basis is provided for multi-source geographic information data, enabling the data to be automatically assigned explicit semantic information, thereby forming enhanced geospatial data.

[0091] As can be seen, this invention achieves the effect of systematically generating geographic information results rich in semantic knowledge from unstructured text and raw data.

[0092] The path exploration module 203 is used to conduct parallel exploration within the constrained space defined by the enhanced geospatial data through the spatial exploration agent, and generate candidate path schemes in real time.

[0093] Specifically, this step enables the spatial exploration agent to conduct parallel exploration within the semantic constraint space, thereby achieving real-time generation of candidate path solutions and improving the efficiency and diversity of path discovery.

[0094] As can be seen, this invention achieves the effect of rapidly generating multiple feasible path alternatives under clear geographical semantic constraints.

[0095] The compliance deduction module 204 is used to synchronously call the specification understanding agent to perform compliance deduction and risk quantification on the path segment in each exploration assessment, and use the results as feedback signals to dynamically adjust the search strategy of the space exploration agent.

[0096] Specifically, by conducting real-time compliance simulations and risk quantifications, and using the results as feedback to dynamically adjust search strategies, the path exploration process can promptly avoid risks and comply with regulatory constraints.

[0097] As can be seen, the present invention achieves the effect of simultaneously completing compliance optimization during the path generation process and guiding the search direction to converge toward a more feasible region.

[0098] The structure generation module 205 is used to automatically generate a three-dimensional structural scheme for the tower and foundation based on the optimized path scheme, combined with terrain and load data, and driven by the design constraints output by the specification understanding agent.

[0099] Specifically, based on the route plan, terrain and load data, and driven by design constraints, the automatic generation of three-dimensional structural schemes for towers and foundations was achieved, improving the automation level of structural design.

[0100] As can be seen, this invention achieves the effect of combining route path, environmental data and design specifications to automatically output a matching three-dimensional structural design scheme.

[0101] The simulation verification module 206 is used to perform simulation verification on the generated structural scheme and feed back the simulation performance data to the specification understanding agent, driving the multi-agent collaborative system to further optimize the scheme that does not meet the performance threshold.

[0102] Specifically, performance data is generated through simulation verification and fed back to the specification understanding agent, driving the system to further optimize solutions that do not meet the threshold, forming a closed-loop iteration based on performance verification.

[0103] As can be seen, the present invention achieves the effect of automatically driving the scheme iteration through simulation feedback data, thereby systematically improving the performance and reliability of the final output scheme.

[0104] Example 3 Please see Figure 3 , Figure 3 This is a schematic diagram of another transmission line scheme generation and optimization system based on multi-agent collaboration disclosed in an embodiment of the present invention. Figure 3 As shown, the device may include: Memory 301 storing executable program code; Processor 302 coupled to memory 301; The processor 302 calls the executable program code stored in the memory 301 to execute some or all of the steps in the method for generating and optimizing transmission line schemes based on multi-agent cooperation disclosed in Embodiment 1 of the present invention.

[0105] Example 4 This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute some or all of the steps in the method for generating and optimizing transmission line schemes based on multi-agent cooperation disclosed in Embodiment 1 of this invention.

[0106] Example 5 This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps of a method for generating and optimizing transmission line schemes based on multi-agent cooperation as described in Embodiment 1.

[0107] The system embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0108] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0109] Finally, it should be noted that the method and system for generating and optimizing transmission line schemes based on multi-agent collaboration disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating and optimizing transmission line schemes based on multi-agent cooperation, characterized in that, The method includes: Construct a multi-agent collaborative system comprising a specification understanding agent, a space exploration agent, and a structure generation agent; The specification understanding agent parses and models the specification text to form a specification knowledge network, and interprets multi-source geographic information data based on this network to output enhanced geospatial data with semantic tags. The spatial exploration agent performs parallel exploration within the constrained space defined by the enhanced geospatial data, generating candidate path schemes in real time; specifically, the spatial exploration agent performs parallel exploration within the semantic constrained space to achieve real-time generation of candidate path schemes. In each exploration and evaluation, the specification understanding agent is synchronously invoked to perform compliance deduction and risk quantification on the path segments, and the results are used as feedback signals to dynamically adjust the search strategy of the space exploration agent; specifically, by performing compliance deduction and risk quantification in real time, and using the results as feedback to dynamically adjust the search strategy, the path exploration process can avoid risks in a timely manner and comply with the specification constraints. The structure generates an intelligent agent that, based on the optimized path scheme and combined with terrain and load data, automatically generates a three-dimensional structural scheme for the tower and foundation under the design constraints output by the specification understanding agent. The generated structural scheme is simulated and verified, and the simulation performance data is fed back to the specification understanding agent to drive the multi-agent cooperative system to further optimize the scheme that does not meet the performance threshold. The process by which the specification-understanding agent performs compliance deduction and risk quantification on path segments specifically involves generating a dynamic compliance energy field. For a point on the path Compliance energy value The calculation formula is: ； in, Point Violation of Article The degree of constraint of class specifications, For the preset nonlinear mapping function, Dynamic weighting coefficients related to the current project context; The space exploration agent employs a multi-objective optimization algorithm, and its fitness function F is constructed as follows: ； in, For path The comprehensive performance index function, For compliant energy fields along the path The path integral is used to deeply embed compliance constraints into the optimization objective by minimizing this integral.

2. The method for generating and optimizing transmission line schemes based on multi-agent collaboration according to claim 1, characterized in that, Before activating the space exploration agent, the specification understanding agent pre-generates a highly compliant guidance channel in the enhanced geospatial data based on core mandatory constraints, as the initial prior information for the space exploration agent.

3. The method for generating and optimizing transmission line schemes based on multi-agent collaboration according to claim 1, characterized in that, The structure generation agent is constructed based on a conditional generation model. Its generation process takes path profile, load conditions, and design constraint latent vectors extracted from the specification understanding agent as conditional inputs, and outputs parameterized structural schemes.

4. The method for generating and optimizing transmission line schemes based on multi-agent collaboration according to claim 1, characterized in that, After generating the structural scheme, the finite element analysis engine is called to perform mechanical verification. If the verification fails, the stress exceeding the standard information and the original design constraints are fed back to the structure generation agent to drive iterative modification of the local structure.

5. The method for generating and optimizing transmission line schemes based on multi-agent collaboration according to claim 1, characterized in that, When the simulation performance data fails to meet the requirements, it is converted into a structured performance specification fragment and added to the input knowledge network of the specification understanding agent to guide the generation of subsequent schemes.

6. The method for generating and optimizing transmission line schemes based on multi-agent collaboration according to claim 1, characterized in that, Also includes: The lightweight versions of the specification understanding agent and the space exploration agent are deployed on mobile terminals, and augmented reality devices are used to visualize and interactively adjust the on-site solutions. The adjusted data is then transmitted back to update the central system model.

7. A transmission line scheme generation and optimization system based on multi-agent cooperation, characterized in that, The system includes: The system construction module is used to build a multi-agent collaborative system that includes a specification understanding agent, a space exploration agent, and a structure generation agent; The specification understanding module is used to parse and model the specification text through the specification understanding agent to form a specification knowledge network, and interpret multi-source geographic information data based on this network to output enhanced geospatial data with semantic tags. The path exploration module is used to perform parallel exploration within the constraint space defined by the enhanced geospatial data through the spatial exploration agent, and generate candidate path schemes in real time; specifically, it achieves real-time generation of candidate path schemes by performing parallel exploration within the semantic constraint space through the spatial exploration agent. The compliance simulation module is used to synchronously invoke the specification understanding agent to perform compliance simulation and risk quantification on the path segments during each exploration assessment, and use the results as feedback signals to dynamically adjust the search strategy of the space exploration agent; specifically, by performing compliance simulation and risk quantification in real time, and using the results as feedback to dynamically adjust the search strategy, the path exploration process can avoid risks in a timely manner and comply with the specification constraints. The structure generation module is used to generate a three-dimensional structural scheme for the tower and foundation based on the optimized path scheme and combined with terrain and load data, driven by the design constraints output by the specification understanding agent. The simulation verification module is used to perform simulation verification on the generated structural scheme, feed back the simulation performance data to the specification understanding agent, and drive the multi-agent collaborative system to further optimize the scheme that does not meet the performance threshold. The process by which the specification-understanding agent performs compliance deduction and risk quantification on path segments specifically involves generating a dynamic compliance energy field. For a point on the path Compliance energy value The calculation formula is: ； in, Point Violation of Article The degree of constraint of class specifications, For the preset nonlinear mapping function, Dynamic weighting coefficients related to the current project context; The space exploration agent employs a multi-objective optimization algorithm, and its fitness function F is constructed as follows: ； in, For path The comprehensive performance index function, For compliant energy fields along the path The path integral is used to deeply embed compliance constraints into the optimization objective by minimizing this integral.

8. A transmission line scheme generation and optimization system based on multi-agent cooperation, characterized in that, The system includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the transmission line scheme generation and optimization method based on multi-agent cooperation as described in any one of claims 1-6.

Images