Power distribution network operation management method fused with large model optimization

Abstract

A power distribution network operation management method is fused with a large model optimization technology and comprises the steps that S1) real-time monitoring, distributed energy operation and historical data are integrated, and data integrity is ensured through preprocessing; s2) combining human expert end-to-end modeling and LLM team dynamic variable generation to form a preliminary power system optimization model; s3) a human expert audits the model generated by the LLM, corrects errors, and improves the accuracy of the model and the safety of the system; and S4) further optimizing and solving by the intelligent decision agent, applying an algorithm library and a solver to obtain an optimal scheme, and displaying a result support decision. A hybrid intelligent decision-making mechanism is introduced into an electric power system, domain knowledge of human experts is combined with the computing power of LLM to form a comprehensive intelligent decision-making support system, the optimization level of the electric power system can be improved, the performance in the aspects of coping with complexity and uncertainty is excellent, and the method is suitable for large-scale popularization and application. The method is superior to a traditional method and a single LLM method in the aspects of intelligence, efficiency and economy.

Description

Technical Field

[0001] The present invention relates to power system operation management technology, and in particular to a distribution network operation management method integrating large model optimization. Background Art

[0002] With the increasing complexity of power systems and the increasing demand for renewable energy, traditional power system decision-making methods can no longer fully meet the needs of modern power systems. Modern power systems not only need to consider traditional power generation and transmission, but also need to integrate distributed energy resources (DER), such as solar power generation, wind power generation, and electric vehicle (EV) charging and discharging management. The introduction of these distributed energy resources increases the complexity and uncertainty of the power system, so a more intelligent and flexible decision support system is needed to achieve optimal management of the power system.

[0003] In order to meet this challenge, the application of large language models (LLM) in power systems has emerged in recent years. LLM is able to process large amounts of natural language data and generate corresponding mathematical models and codes for decision support in power systems. However, current LLMs still have limitations in dealing with complex power system optimization problems and perform poorly in generating complex power flow models and advanced optimization models such as second-order cone relaxation. This results in the low feasibility of relying solely on LLMs for power system optimization decisions.

[0004] At present, the main technical problems facing the power system include the following aspects:

[0005] 1. Complexity and uncertainty: The complexity and uncertainty of the power system have greatly increased, especially when integrating a large number of distributed energy resources. These resources are volatile and intermittent, which increases the complexity of system operation.

[0006] 2. Limitations of traditional methods: Traditional power system optimization methods mainly rely on the knowledge and experience of human experts, which is time-consuming and difficult to cope with dynamically changing system requirements. In addition, traditional methods have obvious deficiencies in processing large-scale data and real-time decision-making.

[0007] 3. Limitations of LLM: Although LLM performs well in natural language processing, it still has obvious limitations in the generation of power system optimization models. For example, LLM has difficulty in generating complex power flow models containing second-order cone relaxation, which limits its application in advanced decision support for power systems.

[0008] 4. Model fusion and decision support: How to effectively integrate the knowledge of human experts and the computing power of LLM to form an efficient and intelligent decision support system is a technical problem that needs to be solved urgently. The system not only needs to have real-time response capabilities, but also needs to be able to handle complex optimization problems to improve the operating efficiency and reliability of the power system.

[0009] It should be noted that the information disclosed in the above background technology section is only used for understanding the background of the present application, and therefore may include information that does not constitute prior art known to ordinary technicians in the field. Summary of the invention

[0010] The main purpose of the present invention is to solve the problems existing in the above-mentioned background technology and to provide a distribution network operation management method integrating large model optimization.

[0011] To achieve the above object, the present invention adopts the following technical solutions:

[0012] A distribution network operation management method integrating large model optimization includes the following steps:

[0013] S1. Data integration and preprocessing: Collect and integrate system real-time monitoring data, distributed energy operation data and historical data; ensure data integrity and consistency through preprocessing modules; use data integration interface to input cleaned data and problem description into the large language model expert group;

[0014] S2, Hybrid Expert Chain Modeling: Human experts are responsible for end-to-end modeling of the power system, including the fixed part of the system topology and distributed generators; the large language model expert team generates dynamic variables and constraints of distributed energy resources based on data and natural language descriptions; coordination experts summarize the results of human experts and LLM experts to form a preliminary power system optimization model;

[0015] S3. Human-in-the-loop model review and correction: LLM generates a preliminary optimization model based on natural language descriptions; human experts conduct a detailed review of the model generated by LLM to identify and correct potential errors and deficiencies; the corrected model ensures accuracy and feasibility, and improves the reliability and security of the system;

[0016] S4. Intelligent decision generation and optimization solution: The power system intelligent decision agent receives the reviewed and revised model and performs further optimization and solution. It applies the appropriate algorithm in the optimization algorithm library and solves the model through the solver to obtain the optimal decision solution. The intelligent decision agent integrates domain knowledge and data, displays the optimization results, and provides decision support for operators.

[0017] A computer-readable storage medium stores a computer program, which implements the method when executed by a processor.

[0018] A computer program product comprises a computer program, which implements the method when executed by a processor.

[0019] The present invention has the following beneficial effects:

[0020] The present invention proposes a hybrid mechanism that integrates human intelligence and LLM, which can fully utilize the potential of LLM while making up for its shortcomings in specific domain knowledge.

[0021] The present invention introduces a hybrid intelligent decision-making mechanism into the power system, combining the domain knowledge of human experts with the computing power of LLM to form a comprehensive intelligent decision support system. This system can not only improve the optimization level of the power system, but also perform well in dealing with complexity and uncertainty.

[0022] Experimental results show that the proposed hybrid intelligent decision-making system is superior to traditional methods and single LLM method in terms of intelligence, efficiency and economy.

[0023] Other beneficial effects of the embodiments of the present invention will be further described below. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a flow chart of the distribution network operation management method optimized by the integrated large model according to an embodiment of the present invention, which is based on the human-in-the-loop hybrid expert chain mechanism framework. DETAILED DESCRIPTION

[0025] The following is a detailed description of the embodiments of the present invention. It should be emphasized that the following description is only exemplary and is not intended to limit the scope and application of the present invention.

[0026] In order to effectively solve the complexity and uncertainty problems existing in the power system, this paper proposes a distribution network operation management method integrating large model optimization to realize a hybrid intelligent decision support system integrating human intelligence and large-scale language model (LLM). Through the "hybrid expert chain" mechanism, the system combines the domain knowledge of human experts with the computing power of LLM to realize intelligent optimization decision-making for the power system.

[0027] See also Figure 1 , a distribution network operation management method integrating large model optimization, comprising the following steps:

[0028] S1. Data integration and preprocessing: Collect and integrate system real-time monitoring data, distributed energy operation data and historical data; ensure data integrity and consistency through preprocessing modules; use data integration interface to input cleaned data and problem description into the large language model expert group;

[0029] S2, Hybrid Expert Chain Modeling: Human experts are responsible for end-to-end modeling of the power system, including the fixed part of the system topology and distributed generators; the large language model expert team generates dynamic variables and constraints of distributed energy resources based on data and natural language descriptions; coordination experts summarize the results of human experts and LLM experts to form a preliminary power system optimization model;

[0030] S3. Human-in-the-loop model review and correction: LLM generates a preliminary optimization model based on natural language descriptions; human experts conduct a detailed review of the model generated by LLM to identify and correct potential errors and deficiencies; the corrected model ensures accuracy and feasibility, and improves the reliability and security of the system;

[0031] S4. Intelligent decision generation and optimization solution: The power system intelligent decision agent receives the reviewed and revised model and performs further optimization and solution. It applies the appropriate algorithm in the optimization algorithm library and solves the model through the solver to obtain the optimal decision solution. The intelligent decision agent integrates domain knowledge and data, displays the optimization results, and provides decision support for operators.

[0032] In some embodiments, in step S2, hybrid expert chain modeling includes:

[0033] S2.1. Clearly define the specific tasks and responsibilities of each expert role in the human experts and LLM expert team, where the human experts are responsible for the end-to-end modeling of the power system, including but not limited to the construction of the system topology and the parameter setting of distributed generators; the LLM expert team is responsible for generating dynamic variables and constraints of distributed energy resources based on data and natural language descriptions;

[0034] S2.2, the coordination experts assign tasks to the corresponding human experts and LLM experts according to the specific needs of power system optimization, and supervise the execution of the tasks;

[0035] S2.3, each expert in the LLM expert team generates dynamic variables and constraints of distributed energy resources according to their professional fields;

[0036] S2.4. The coordination experts in the LLM expert team integrate the models and variables generated by human experts and the LLM expert team, and optimize the model structure through specific algorithms and methods to ensure the computational efficiency of the model and the accuracy of the solution;

[0037] S2.5. Based on the hybrid expert chain mechanism, a preliminary power system optimization decision model is formed, which can reflect the actual operating status and demand of the power system and provide a basis for subsequent review and correction.

[0038] In some embodiments, the LLM expert team includes: simplification experts: responsible for decomposing complex problems into sub-problems that are easier to handle; power system modeling experts: responsible for building mathematical models of power systems; coding experts: responsible for converting models into executable codes; parameter extraction experts: responsible for extracting key parameters from data for use by the model; code review experts: responsible for reviewing the generated code to ensure its accuracy and reliability; and coordination experts: summarizing the opinions of various experts and ultimately generating an executable optimization model.

[0039] In some embodiments, in step S3, the human-in-the-loop model review and correction specifically includes:

[0040] S3.1, LLM receives the power system problem described in natural language and uses its deep learning algorithm to generate a preliminary optimization model and code;

[0041] S3.2, human experts review the preliminary model generated by LLM to ensure that the model meets the actual operation requirements of the power system;

[0042] S3.3, human experts identify and correct potential errors and deficiencies in the LLM generation model;

[0043] S3.4. Verify the accuracy of the model by comparing and analyzing the actual power system data to ensure that the revised model accurately reflects the actual operation of the power system;

[0044] S3.5. Improve the safety and reliability of the model through iteration and verification;

[0045] S3.6. Feedback the review and correction results of human experts to LLM to facilitate self-learning and optimization of the model, reduce errors in future model generation, and improve the overall quality of model generation.

[0046] In some embodiments, step S3 includes:

[0047] LLM uses natural language processing technology and deep learning algorithms to generate dynamic variables and constraints of distributed energy resources based on natural language descriptions;

[0048] Human experts evaluate the models generated by LLM through a series of standardized review steps, including the accuracy and completeness of the structure, parameters, and constraints;

[0049] Based on the feedback from human experts, LLM makes model revisions and optimizes the model through an iterative process until the required accuracy and feasibility are achieved;

[0050] The revised model can be verified by simulation, historical data analysis and other technical means to ensure its accuracy and feasibility in actual power system operation.

[0051] In some embodiments, in step S4, intelligent decision making and optimization solution specifically include:

[0052] The input interface of the intelligent decision-making agent obtains data from different sources, including real-time monitoring data, distributed energy operation data and historical data, and ensures the availability and accuracy of the data through preprocessing;

[0053] Based on the integrated domain knowledge and data, the intelligent decision agent generates a hybrid optimization model to reflect the actual operating status and constraints of the power system;

[0054] The optimization and learning process of the intelligent decision-making agent selects appropriate optimization algorithms to adapt to different optimization scenarios and requirements;

[0055] The generated optimization model is solved using a solver to obtain the optimal or approximately optimal decision solution.

[0056] In some embodiments, the method further includes: S5, verifying the effectiveness of the model through actual data and simulation experiments, and adjusting and optimizing the model according to the experimental results to ensure its reliability and efficiency in practical applications.

[0057] The present invention introduces a hybrid intelligent decision-making mechanism into the power system, combining the domain knowledge of human experts with the computing power of LLM to form a comprehensive intelligent decision support system, which can not only improve the optimization level of the power system, but also perform well in dealing with complexity and uncertainty. It is superior to traditional methods and a single LLM method in terms of intelligence, efficiency and economy.

[0058] The specific embodiments of the present invention are further described below.

[0059] Hybrid expert chain mechanism: The embodiment of the present invention proposes a "hybrid expert chain" mechanism that combines human experts and LLM expert teams to achieve efficient power system decision-making. Human experts are responsible for end-to-end modeling of the power system, including the system topology and the fixed part of distributed generators; the LLM expert team generates dynamic variables and constraints of distributed energy resources through specific prompts, and finally integrates the results of the two to obtain the final power system model. This mechanism effectively integrates the experience of human experts and the computational advantages of LLM, and obtains a fusion model that can be directly calculated and processed.

[0060] Human-in-the-loop design: In order to solve the limitations of LLM in generating complex power system models, the embodiment of the present invention introduces "human-in-the-loop" design. The specific operation process is: LLM generates a preliminary optimization model and code based on the natural language description, and then human experts review and revise it to ensure the accuracy and feasibility of the model. This design not only improves the accuracy of the model, but also enhances the reliability and security of the system.

[0061] Power system intelligent decision agent: An LLM-based power system intelligent decision agent is designed, which can process various inputs, generate hybrid optimization models, solve them through open source and commercial solvers, and output the final intelligent decision. The intelligent agent consists of three parts: input interface, optimization and learning process, and output interface, which are used to integrate domain knowledge, generate and solve optimization models, and display results.

[0062] Expert team structure design: In the LLM part, a team structure consisting of multiple LLM experts with different tasks is proposed. The team includes coordination experts, simplification experts, power system modeling experts, coding experts, parameter extraction experts, and code review experts. Each expert generates the corresponding mathematical model and code according to the specific task, and then the coordination expert summarizes the opinions of each expert and finally generates an executable optimization model.

[0063] In view of the above content, a large number of experiments have been conducted to verify the effectiveness and efficiency of the model proposed in this invention. The experimental results show that the proposed hybrid intelligent decision-making system is superior to the traditional method and the single LLM method in terms of intelligence, efficiency and economy.

[0064] 1. In the data input phase, the system real-time monitoring data, distributed energy operation data, historical data, etc. are integrated using the preprocessing module to ensure the integrity and consistency of the data, and are transmitted to the large language model expert group together with the problem description and code framework.

[0065] 2. In the LLM preliminary model generation stage, a large-scale language model (such as GPT-4) is used to generate preliminary optimization models and codes based on data and natural language descriptions.

[0066] 3. Human expert review and correction stage: human experts are used to review and correct the preliminary model generated by LLM to ensure the accuracy and feasibility of the model.

[0067] 4. Optimization algorithm selection and solution stage: According to the specific optimization requirements, select the appropriate optimization algorithm and optimization solver (such as Gurobi, Cplex, etc.) to solve the model and obtain the optimization results. Obtain information such as the operating status of the power system and the utilization of distributed energy.

[0068] The specific implementation process includes:

[0069] 1. Demand analysis and preliminary design: By analyzing the actual needs of the power system, determine the functions and structure of the hybrid intelligent decision-making system, including the division of labor and cooperation mode of human experts and LLM expert teams.

[0070] 2. LLM expert team training: Use large-scale language models (such as GPT-4) to train power system optimization problems in various fields to form a professional LLM expert team. Each expert generates corresponding optimization models and code snippets based on their tasks.

[0071] 3. Hybrid model generation: After LLM experts generate the preliminary distributed energy resource model, human experts review and revise it to ensure the accuracy of the model. Finally, the coordination experts merge the various models to generate a complete power system optimization model.

[0072] 4. Solving and Optimization: Use open source or commercial optimization solvers (such as Gurobi, Cplex) to solve the hybrid optimization model and obtain the optimal decision solution.

[0073] 5. Result verification and adjustment: Verify the effectiveness of the model through actual data and simulation experiments, and adjust and optimize the model according to the experimental results to ensure its reliability and efficiency in practical applications.

[0074] Through the above technical means and methods, the present invention successfully solves the complexity and uncertainty problems in power system optimization decision-making and realizes efficient and intelligent power system management.

[0075] The power system intelligent decision-making agent in the embodiment of the present invention produces remarkable technical effects by combining the domain knowledge of human experts with the computing power of large-scale language models (LLM):

[0076] 1. Improved accuracy and reliability of decision-making: By introducing human-in-the-loop design, intelligent decision-making agents can ensure the accuracy and reliability of the generated optimization models in practical applications. Human experts review and correct the models generated by LLM, which makes up for the lack of knowledge in the field of power systems of LLM and avoids potential safety hazards.

[0077] 2. Enhanced system intelligence and adaptability: Intelligent decision agents can flexibly adjust optimization models and decision-making plans based on real-time data and dynamically changing system requirements. This adaptive capability enables the power system to more effectively cope with the volatility and uncertainty of distributed energy resources, improving system stability and operational efficiency.

[0078] 3. Improved optimization efficiency and economy: By using LLM to generate dynamic variables and constraints of distributed energy resources, the intelligent decision agent significantly reduces the time for model generation and optimization. Compared with traditional methods, the system can complete the generation and solution of complex power system optimization models in a few minutes, greatly improving the optimization efficiency. In addition, the intelligent decision agent can more accurately optimize the operating cost of the power system and reduce the overall operating cost of the power system.

[0079] 4. Automated and efficient decision support: Intelligent decision agents achieve full process automation from data input, model generation to optimization solution through automated processes. This efficient decision support system not only reduces manual intervention and errors, but also improves the overall efficiency and reliability of power system operations.

[0080] 5. Improved model interpretability and transparency: Intelligent decision agents use models and codes generated by LLM, combined with the review and correction of human experts, to ensure the interpretability and transparency of optimization models. This enables power system operators to clearly understand and explain the basis and process of optimization decisions, enhancing control and trust in system operation.

[0081] Through the above technical effects, the power system intelligent decision-making agent of the present invention performs well in improving the intelligence, efficiency and economy of power system optimization decisions, and significantly enhances the stability and operating benefits of the power system.

[0082] The embodiment of the present invention proposes the following innovative designs:

[0083] Hybrid expert chain mechanism:

[0084] A hybrid expert chain mechanism combines human experts and large-scale language model (LLM) expert teams to achieve efficient power system decision-making. The mechanism includes multiple expert roles, each responsible for different tasks, and unified dispatch through coordinating experts.

[0085] a. Cooperation model between human experts and LLM expert team: This mechanism defines the cooperation model between human experts and LLM expert team in detail, and clarifies their respective tasks and responsibilities. For example, human experts are responsible for the overall modeling and complex decision-making of the power system, while the LLM expert team is responsible for generating specific variables and constraints.

[0086] b. Coordination expert scheduling and integration method: Coordination experts play a scheduling and integration role in the entire mechanism to ensure that the tasks of each expert can be carried out in an orderly manner. This method includes how to select appropriate experts to participate in the task and how to combine the opinions of various experts to form the final decision.

[0087] 2. Human-in-the-Loop Design:

[0088] In human-in-the-loop design, LLM generates preliminary optimization models and codes, which are reviewed and revised by human experts to ensure the accuracy, reliability and security of the models.

[0089] a. Preliminary design method of LLM generation model: Describe in detail how LLM generates a preliminary optimization model based on natural language description, including what algorithms and technologies are used to implement this process.

[0090] b. Human expert review and correction process: Clarify the specific steps in the process of human experts reviewing and correcting the model, including how to identify and correct errors in the model, how to ensure the accuracy of the model, etc.

[0091] c. Technical means to ensure the security and reliability of the model: Improve the reliability and security of the model through multiple iterations and verifications, including what verification methods and technical means to use.

[0092] 3. Intelligent decision-making agent for power system:

[0093] An LLM-based intelligent decision-making agent for power systems is designed, which can process various inputs, generate hybrid optimization models, and solve them through open source and commercial solvers to output the final intelligent decision. The intelligent agent consists of three parts: input interface, optimization and learning process, and output interface.

[0094] a. How the input interface integrates domain knowledge: Describes how the input interface obtains and integrates data from different sources, including which data sources and integration methods are used.

[0095] b. Specific algorithms and implementations of the optimization and learning process: Describe in detail the specific algorithms and implementation methods of the optimization and learning process, including which optimization algorithms to use, how to train and adjust the model, etc.

[0096] c. Methods and techniques for displaying results on the output interface: Describe how to display complex optimization results to users in an easy-to-understand form, including what display methods and technical means are used.

[0097] 4.LLM expert team structure:

[0098] A team structure consisting of multiple LLM experts with different tasks is proposed. The team includes coordination experts, simplification experts, power system modeling experts, coding experts, parameter extraction experts, and code review experts.

[0099] a. Definition of specific tasks and responsibilities of each LLM expert: Define in detail the specific tasks and responsibilities of each LLM expert. For example, simplification experts are responsible for simplifying complex problems, coding experts are responsible for generating executable code, etc.

[0100] b. Methods for coordinating experts to integrate the opinions of various experts: Describes how coordinating experts schedule the tasks of various experts and integrate the opinions of various experts to form the final decision.

[0101] c. The process of how the LLM expert team generates optimization models and codes: Describe in detail how the LLM expert team generates optimization models and codes, including the specific operation steps and methods of each expert.

[0102] 5. Optimized model generation method combining LLM generation and human review:

[0103] The dynamic variables and constraints of distributed energy resources are generated through LLM and reviewed and revised by human experts to ensure the accuracy and feasibility of the optimization model.

[0104] a. Methods of generating dynamic variables and constraints by LLM: Describe in detail how LLM generates dynamic variables and constraints of distributed energy resources based on natural language descriptions, including which algorithms and technologies are used.

[0105] b. Specific process of human expert review and correction: Describe how human experts review and correct the model generated by LLM, including the specific steps and methods of the review.

[0106] c. Technical means to ensure model accuracy and feasibility: Improve the accuracy and feasibility of the model through multiple iterations and verifications, including what verification methods and technical means to use.

[0107] 6. Efficient optimization solution and automated decision support system:

[0108] Through the automated process, the whole process from data input, model generation to optimization solution is automated. The system can complete the generation and solution of complex power system optimization models within a few minutes.

[0109] a. Specific implementation method of the automated process: Describe the specific implementation method of the automated process, including the specific steps of data input, model generation and optimization solution.

[0110] b. Use of optimization solving algorithms and tools: Describe in detail the algorithms and tools used in the optimization solving process, including which optimization algorithms and solvers are used.

[0111] c. Technical means to improve optimization efficiency and economy: Describe the methods to improve optimization efficiency and economy through automated processes and efficient algorithms, including specific technical means and implementation methods.

[0112] The method of the present invention includes various aspects from hybrid expert chain mechanism design to specific implementation in power system optimization decision-making, is applied to the field of power system optimization decision-making, and has high technical advancement and market competitiveness.

[0113] An embodiment of the present invention further provides a storage medium for storing a computer program, which at least performs the above method when executed.

[0114] An embodiment of the present invention further provides a control device, comprising a processor and a storage medium for storing a computer program; wherein the processor is configured to execute at least the method described above when executing the computer program.

[0115] An embodiment of the present invention further provides a processor, wherein the processor executes a computer program and at least executes the method described above.

[0116] The storage medium can be implemented by any type of non-volatile storage device, or a combination thereof. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a ferromagnetic random access memory (Flash Memory), a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM); the magnetic surface memory can be a disk memory or a tape memory. The storage medium described in the embodiments of the present invention is intended to include, but is not limited to, these and any other suitable types of memory.

[0117] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0118] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0119] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0120] A person skilled in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, etc. Various media that can store program codes.

[0121] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0122] The methods disclosed in the several method embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0123] The features disclosed in several product embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new product embodiments.

[0124] The features disclosed in several method or device embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0125] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art of the present invention, several equivalent substitutions or obvious variations can be made without departing from the concept of the present invention, and the performance or use is the same, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A distribution network operation management method integrating large model optimization, characterized in that: The steps include: S1. Collect and integrate system real-time monitoring data, distributed energy operation data and historical data; ensure data integrity and consistency through preprocessing modules; use data integration interface to input cleaned data and problem description into large language model expert group; S2. Human experts are responsible for the end-to-end modeling of the power system, including the fixed part of the system topology and distributed generators; the large language model expert team generates dynamic variables and constraints of distributed energy resources based on data and natural language descriptions; coordination experts summarize the results of human experts and LLM experts to form a preliminary power system optimization model; S3, LLM generates a preliminary optimization model based on the natural language description; Human experts conduct detailed reviews of the models generated by LLM to identify and correct potential errors and deficiencies; S4. The power system intelligent decision agent receives the reviewed and revised model and performs further optimization and solution. It applies the appropriate algorithm in the optimization algorithm library and solves the model through the solver to obtain the optimal decision solution. The intelligent decision agent integrates domain knowledge and data, displays the optimization results, and provides decision support for operators.

2. The distribution network operation management method integrating large-scale model optimization as claimed in claim 1 is characterized in that: In step S2, hybrid expert chain modeling includes: S2.

1. Clearly define the specific tasks and responsibilities of each expert role in the human expert and LLM expert team, where the human expert is responsible for the end-to-end modeling of the power system, including the construction of the system topology and the parameter setting of distributed generators; the LLM expert team is responsible for generating dynamic variables and constraints of distributed energy resources based on data and natural language descriptions; S2.2, the coordination experts assign tasks to the corresponding human experts and LLM experts according to the specific needs of power system optimization, and supervise the execution of the tasks; S2.3, each expert in the LLM expert team generates dynamic variables and constraints of distributed energy resources according to their professional fields; S2.4, the coordination experts in the LLM expert team merge the models and variables generated by human experts and the LLM expert team to optimize the model structure; S2.

5. Based on the hybrid expert chain mechanism, a preliminary power system optimization decision model is formed.

3. The distribution network operation management method based on large-scale model optimization as claimed in claim 2 is characterized in that: The LLM expert team includes: Simplification experts: responsible for breaking down complex problems into more manageable sub-problems; Power system modeling expert: responsible for building mathematical models of power systems; Coding experts: responsible for converting models into executable code; Parameter extraction expert: responsible for extracting key parameters from the data for use in the model; Code review experts: responsible for reviewing the generated code to ensure its accuracy and reliability; and Coordinate experts: Summarize the opinions of experts and ultimately generate an executable optimization model.

4. The distribution network operation management method based on large-scale model optimization according to any one of claims 1 to 3, characterized in that: In step S3, the human-in-the-loop model review and correction specifically includes: S3.1, LLM receives the power system problem described in natural language and uses its deep learning algorithm to generate a preliminary optimization model and code; S3.2, human experts review the preliminary model generated by LLM to ensure that the model meets the actual operation requirements of the power system; S3.3, human experts identify and correct potential errors and deficiencies in the LLM generation model; S3.

4. Verify the accuracy of the model by comparing and analyzing the actual power system data to ensure that the revised model accurately reflects the actual operation of the power system; S3.

5. Improve the safety and reliability of the model through iteration and verification; S3.

6. Feedback the review and correction results of human experts to LLM to facilitate self-learning and optimization of the model.

5. The distribution network operation management method based on large-scale model optimization according to any one of claims 1 to 3, characterized in that: Step S3 also includes: LLM uses natural language processing technology and deep learning algorithms to generate dynamic variables and constraints of distributed energy resources based on natural language descriptions; Human experts evaluate the models generated by LLM through a series of standardized review steps, including the accuracy and completeness of the structure, parameters, and constraints; Based on the feedback from human experts, LLM makes model revisions and optimizes the model through an iterative process until the required accuracy and feasibility are achieved; The revised model is verified to ensure its accuracy and feasibility in actual power system operation.

6. The distribution network operation management method integrating large-scale model optimization according to any one of claims 1 to 5, characterized in that: In step S4, intelligent decision making and optimization solution specifically include: The input interface of the intelligent decision-making agent obtains data from different sources, including real-time monitoring data, distributed energy operation data and historical data, and ensures the availability and accuracy of the data through preprocessing; Based on the integrated domain knowledge and data, the intelligent decision agent generates a hybrid optimization model to reflect the actual operating status and constraints of the power system; The optimization and learning process of the intelligent decision-making agent selects appropriate optimization algorithms to adapt to different optimization scenarios and requirements; The generated optimization model is solved using a solver to obtain the optimal or approximately optimal decision solution.

7. The distribution network operation management method integrating large-scale model optimization according to any one of claims 1 to 5, characterized in that: The method further comprises: S5. Verify the effectiveness of the model through actual data and simulation experiments, and adjust and optimize the model based on the experimental results to ensure its reliability and efficiency in practical applications.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images