Comprehensive energy system flexible optimization scheduling method based on large language model

Abstract

The invention discloses an integrated energy system flexible optimization scheduling method based on a large language model. The method comprises the following steps: S1, constructing a multi-source heterogeneous knowledge base and a rule logic coding system of the comprehensive energy system; s2, training a large language model agent with a dynamic reasoning capability in the field of the integrated energy system; s3, analyzing a user input demand, and constructing an optimal scheduling scene comprising adjustable equipment, a scheduling time period and a target weight; s4, dynamically predicting the carbon price and the energy price in the scheduling time period; s5, establishing a flexible optimization scheduling model containing a target function and constraint conditions according to the optimization scheduling scene; and S6, solving the flexible optimization scheduling model, updating model cognition in real time according to user feedback and a scheduling result, and realizing adaptive optimization of a scheduling strategy. According to the invention, the flexibility and adaptability of the optimization scheduling process can be effectively improved, and flexible, efficient and low-carbon operation of the integrated energy system is facilitated.

Description

Technical Field

[0001] The present invention belongs to the field of optimized scheduling of integrated energy systems, and in particular relates to a flexible optimized scheduling method for integrated energy systems based on a large language model. Background Art

[0002] Driven by the global energy transition, integrated energy systems, by integrating multiple energy sources such as electricity, gas, and heat, have become a key vehicle for improving energy efficiency and promoting the absorption of renewable energy. Integrated energy systems facilitate the complementary and cascaded utilization of multiple energy sources, such as using industrial waste heat for power generation and combining photovoltaics with energy storage to power buildings. With the rapid increase in the proportion of fluctuating power sources such as wind power and photovoltaics, and the growing demand for energy cost control in factories and industrial parks, the large-scale application of integrated energy systems has become a trend. Currently, over 60% of industrial parks in my country have gradually introduced integrated energy services, and their low-carbon and flexible operating models provide key support for the transformation of the energy structure.

[0003] Optimal scheduling of integrated energy systems is a core component of energy conservation and carbon reduction. By coordinating the output of various energy devices (such as gas turbines, energy storage systems, and heat pumps), optimized scheduling can significantly reduce system operating costs while also reducing fossil energy consumption and carbon emissions. For example, storing electricity during periods of low electricity prices and discharging it during peak periods, or prioritizing low-carbon equipment when carbon prices rise, can achieve a win-win situation for both economic and environmental benefits. However, with increasing system complexity, improvements in carbon market mechanisms, and reforms to the electricity spot market, energy prices and policy constraints are fluctuating increasingly frequently. Traditional static optimization strategies struggle to adapt to the dynamic changes in system states and market conditions, necessitating a scheduling method that can respond in real time and flexibly adjust.

[0004] The current integrated energy system optimization and scheduling methods still have obvious limitations, including: (1) Most optimization models use fixed mathematical formulas and preset conditions. Once users need to adjust the scheduling time scale (such as changing from day-ahead scheduling to intraday real-time optimization) or add or remove equipment (such as temporarily connecting energy storage equipment), they often need to rebuild the model, which is time-consuming and labor-intensive; (2) Existing methods usually rely on historical data for parameter setting, which makes it difficult to integrate real-time market fluctuations, policy adjustments or equipment status changes, resulting in the scheduling strategy gradually becoming out of touch with actual needs during use. Summary of the Invention

[0005] In order to solve the problems existing in the existing energy system optimization and scheduling methods, the present invention provides a flexible optimization and scheduling method for an integrated energy system based on a large language model. This method can create flexible optimization and scheduling scenarios according to user needs and continuously update intelligent agent cognition, thereby improving the flexibility and adaptability of the optimization and scheduling process, which will help my country build a new type of efficient and low-carbon energy system.

[0006] The present invention is achieved by adopting the following technical solutions:

[0007] A flexible optimization scheduling method for an integrated energy system based on a large language model includes the following steps:

[0008] Step S1: Constructing a multi-source heterogeneous knowledge base and rule logic coding system for the integrated energy system. Step S1 specifically includes the following steps:

[0009] Step S11, constructing a multi-source heterogeneous knowledge base for an integrated energy system; specifically: integrating the technical parameters of integrated energy system equipment, system topology, transaction mechanism rules, policy texts, equipment and load historical data, historical energy prices and carbon prices, and uniformly storing and associating structured data (specifically including technical parameters of integrated energy system equipment, system topology, equipment and load historical data, historical energy prices and carbon prices) with unstructured data (specifically including transaction mechanism rules and policy texts), and establishing a multi-source heterogeneous knowledge base for an integrated energy system, wherein the multi-source heterogeneous knowledge base for an integrated energy system supports semantic retrieval. Semantic retrieval specifically refers to realizing intelligent association matching between user statements and data stored in the knowledge base by parsing the contextual semantics of user natural language input statements, and returning data that matches the semantic meaning. Figure 1 consistent structured parameters rather than relying on traditional keyword matching.

[0010] Step S12 constructs a rule logic coding system. Specifically, it performs semantic parsing and entity relationship extraction on the unstructured data in the multi-source heterogeneous knowledge base of the integrated energy system, identifies the operating rules, constraints, and policy clauses in the domain knowledge of the integrated energy system, and converts them into computer-executable formal logical expressions (specifically, standardized expressions formed by mathematical symbols and logical operators that can be directly called and executed by computer programs). A dynamic mapping mechanism is used to bind the operating rules to the system's real-time state variables to generate a computable constraint set that adapts to different scheduling scenarios (the constraints in the constraint set use computer-executable formal logical expressions). This computable constraint set is then integrated with the structured data in the multi-source heterogeneous knowledge base of the integrated energy system to obtain the rule logic coding system. The rule logic coding system supports multi-dimensional logical reasoning. Multi-dimensional logical reasoning refers to establishing association and priority mechanisms between rules of different dimensions, thereby conducting multi-angle analysis and weighing different possibilities under complex scheduling instructions, ultimately generating coherent and reasonable conclusions.

[0011] Step S2: Based on the multi-source heterogeneous knowledge base and rule logic coding system of the integrated energy system, a large language model agent with dynamic reasoning capabilities in the field of integrated energy systems is trained. Step S2 is specifically as follows:

[0012] Step S21, pre-training of basic capabilities; specifically: based on the pre-trained language model base, using the multi-source heterogeneous knowledge base of the integrated energy system, through supervised learning, a mapping relationship between natural language instructions and structured parameters such as adjustable equipment, scheduling time periods, and scheduling target weights is established to obtain a large language model with the ability to semantically understand the terminology and basic rules in the field of the integrated energy system; and based on the rule logic coding system, the semantic understanding ability of the large language model for the terminology and basic rules in the field of the integrated energy system is further strengthened.

[0013] Step S22: Enhanced training of complex reasoning capabilities; specifically: through a multi-task joint training mechanism, the large language model is enabled to synchronously learn semantic parsing, temporal trend reasoning, and dynamic constraint generation capabilities (specifically, the ability to dynamically adjust or generate new computable constraints based on user semantics); at the same time, the large language model derives mathematical constraints that match the current scheduling scenario based on real-time device status, user demand priority, and rule logic coding system, and calls the computable constraint set in the rule logic coding system to verify logical consistency, thereby improving the complex reasoning capabilities of the large language model, and ultimately enabling the large language model to achieve end-to-end reasoning from fuzzy requirements to precise scheduling parameters.

[0014] Step S23, incremental learning interface deployment; specifically: the large language model dynamically accesses equipment change parameters, policy change texts and user feedback information, and realizes continuous iteration of the large language model's cognitive ability through efficient parameter fine-tuning methods, and synchronously updates the rules and associations of the constraints in the rule logic coding system (rules refer to the mandatory restrictions of constraints, such as the upper and lower limits of equipment power; associations refer to the logical relationships between constraints, such as whether they conflict with each other), ensuring the long-term adaptability of the large language model's reasoning ability and the dynamic evolution of the system, thereby obtaining a large language model intelligent entity with dynamic reasoning capabilities in the field of integrated energy systems.

[0015] Step S3: Analyze the user input using the large language model agent built in step S2 to obtain the integrated energy system scheduling requirements, which include adjustable equipment, scheduling period, and scheduling target weight; and construct an optimized scheduling scenario based on the integrated energy system scheduling requirements. Step S3 is specifically as follows:

[0016] Based on the natural language understanding capability of the large language model, the text scheduling requirements entered by the user are converted into integrated energy system scheduling requirements. If the user-input requirements do not mention the specified content, they are set as the default value when the large language model intelligent agent is constructed.

[0017] An optimized scheduling scenario is constructed based on the scheduling requirements of the integrated energy system, which is specifically expressed as follows:

[0018] Adjustable device collection for:

[0019]

[0020] in, A collection of all devices in the system. Representation device In the real-time state at time t, if the state is 1, it is adjustable.

[0021] The scheduling period T is:

[0022]

[0023] in, To schedule the time, is the scheduling start time, , k is the number of scheduling times, a total of N scheduling times.

[0024] Scheduling target weight parameter for:

[0025]

[0026] in, is the energy cost weight, is the carbon emission cost weight.

[0027] Step S4: Use the large language model agent to dynamically predict the carbon price and energy price within the scheduling period. Specifically:

[0028] Based on a large language model agent, this system uses historical price data and user input to generate carbon price and energy price forecast sequences for the dispatch period. User input includes real-time policy text and supply and demand situation information. If the user does not provide specific information, the forecast sequence is generated using the default parameters of the large language model agent.

[0029] The carbon price prediction series for:

[0030]

[0031] in, is the time series trend analysis function, is the historical carbon price data, The carbon price weight coefficient for user input information, is the semantic influence quantification function, Enter information for the user.

[0032] The energy price forecast series for:

[0033]

[0034] Where m is the energy type used by the equipment (including gas, steam, electricity), is the historical price data of energy of the corresponding type, The energy price weight coefficient for the corresponding type in the user input information.

[0035] Step S5: Based on the optimized scheduling scenario and the predicted carbon and energy prices, a flexible optimization scheduling model including objective functions and constraints is established. Specifically:

[0036] Based on the parameters in the optimized scheduling scenario obtained in step S3 and the carbon price and energy price forecast sequence obtained in step S4, a flexible optimized scheduling model is constructed. The flexible optimized scheduling model uses the minimization of the weighted sum of carbon cost and energy cost as the objective function, and uses equipment output range constraints, equipment output ramp constraints, multi-energy flow balance constraints, and system topology constraints as constraints. Solving this model can obtain the optimal output strategy for adjustable equipment within the scheduling period. By parsing user input requirements into integrated energy system scheduling requirements, an optimized scheduling scenario is constructed, and a carbon price and energy price forecast sequence is dynamically predicted. The established objective function and constraints can be flexibly changed to dynamically meet actual scheduling needs.

[0037] The objective function is:

[0038]

[0039] in, is the objective function value, For device i Output of the time period, device i belongs to the set of adjustable devices , is the carbon emission coefficient of device i; the remaining parameters in the objective function are flexibly generated through the aforementioned steps and after parsing the user input.

[0040] The constraint condition parameter set is included in the rule logic coding system when the large language model is constructed, and the flexible optimization scheduling model selects a constraint condition parameter subset corresponding to the adjustable device set from it.

[0041] Step S6: solve the flexible optimization scheduling model, and update the large language model agent of the model in real time according to user feedback and scheduling results, and realize adaptive optimization of the scheduling strategy based on the large language model agent. Specifically:

[0042] Solve the flexible optimization scheduling model to obtain the optimal output of each adjustable device at each scheduling moment. Based on the user's further correction instructions for the obtained scheduling strategy and actual operating data, the large language model agent parses the correction instructions and extracts the integrated energy system scheduling requirements. It dynamically updates the parameters of the flexible optimization scheduling model and regenerates the scheduling strategy that meets the user's latest needs.

[0043] At the same time, the operation deviation data is obtained based on the correction instructions and actual operation data, and the correction instructions and operation deviation data are input into the large language model intelligent agent. The price prediction logic of step S4 and the scheduling demand analysis rules of step S3 are updated through the incremental learning mechanism, and the multi-source heterogeneous knowledge base of the integrated energy system is optimized simultaneously to ensure that the subsequent scheduling strategy can adapt to external changes and ensure the effectiveness and robustness of the long-term scheduling effect.

[0044] Finally, adaptive optimization of the scheduling strategy can be achieved based on the large language model agent.

[0045] The present invention also provides a computer-readable storage medium having computer instructions stored thereon, wherein the computer instructions are used to enable a computer to execute the flexible optimization scheduling method for an integrated energy system based on a large language model.

[0046] The beneficial effects of the present invention are:

[0047] The present invention achieves dynamic coupling of scheduling requirements and system status through a large language model intelligent agent, breaking through the limitations of traditional optimization methods that rely on fixed models and static parameters. On the one hand, based on natural language interaction, it supports users to flexibly adjust the scheduling time scale, scheduling target weight and adjustable equipment range, and can quickly generate adaptive optimization scenarios without reconstructing the model; on the other hand, through incremental learning of the large language model, it ensures the synchronous evolution of scheduling strategies and environmental changes, avoiding the problem of scheduling strategy failure. The method of the present invention can be beneficial to improving the flexibility and adaptability of the optimization scheduling strategy of the integrated energy system, and provides technical support for building a new energy system with high efficiency and low carbon. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is the main step of the present invention;

[0049] Figure 2 Schematic diagram of the mechanism for flexible optimization and scheduling of the integrated energy system based on the large language model intelligent agent described in the present invention. DETAILED DESCRIPTION

[0050] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0051] Example 1

[0052] like Figure 1 As shown, the present invention provides a flexible optimization scheduling method for an integrated energy system based on a large language model, the main steps of which include:

[0053] In step S1:

[0054] Step S11, constructing a multi-source heterogeneous knowledge base of the integrated energy system; collecting equipment technical parameters (including rated power and efficiency), system topology (stored in the form of an adjacency matrix), electricity market transaction rule documents and policy documents (PDF / text format, unstructured data, all of which are structured data), equipment and load historical data (CSV structured data), historical energy prices and carbon prices (CSV structured data) from the SCADA system, equipment manufacturer database and policy release platform of the integrated energy system; and uniformly storing and associating the structured data with the unstructured data to establish a multi-source heterogeneous knowledge base of the integrated energy system that supports semantic retrieval.

[0055] Step S12, constructing a rule logic coding system; specifically: performing semantic analysis and entity relationship extraction on the unstructured data in the multi-source heterogeneous knowledge base of the integrated energy system, identifying the operating rules, constraints and policy clauses in the domain knowledge of the integrated energy system, and converting them into computer-executable IF-THEN logical statements or JSON format files; binding the operating rules to the real-time state variables of the system through a dynamic mapping mechanism, generating a computable constraint set that adapts to different scheduling scenarios, and integrating the computable constraint set with the structured data in the multi-source heterogeneous knowledge base of the integrated energy system through the API interface, and constructing a rule logic coding system that supports multi-dimensional logical reasoning.

[0056] In step S2:

[0057] Step S21, pre-training of basic capabilities; the specific method is: using the pre-trained language model base of the Transformer architecture, utilizing the multi-source heterogeneous knowledge base of the integrated energy system, establishing a mapping relationship between natural language instructions and structured parameters such as adjustable equipment, scheduling periods, and scheduling target weights through supervised learning, and further strengthening the semantic understanding ability of the large language model for energy system field terminology and basic rules based on the rule-based logic coding system.

[0058] Step S22: Strengthening training of complex reasoning capabilities. The specific method is: through a multi-task joint training mechanism, the large language model is enabled to synchronously learn semantic parsing, temporal trend reasoning, and dynamic constraint generation capabilities. At the same time, the large language model is based on real-time device status, user demand priority, and rule logic coding system to derive mathematical constraints that match the current scheduling scenario, and call the computable constraint set to verify logical consistency, thereby improving the complex reasoning capabilities of the large language model, and ultimately enabling the large language model to achieve end-to-end reasoning from fuzzy requirements to precise scheduling parameters.

[0059] Step S23, incremental learning interface deployment; the specific method is: the large language model dynamically accesses equipment change parameters, policy change texts and user feedback information, and realizes the continuous iteration of the cognitive ability of the large language model through an efficient parameter fine-tuning method, and synchronously updates the rules and associations of the constraints in the rule logic coding system to ensure the long-term adaptability of the large language model's reasoning ability and the dynamic evolution of the system, thereby obtaining a large language model intelligent entity with dynamic reasoning capabilities in the field of integrated energy systems.

[0060] In step S3:

[0061] Based on the natural language understanding capabilities of the large language model, the text scheduling requirements entered by the user are converted into integrated energy system scheduling requirements, specifically including adjustable equipment, scheduling period, and scheduling target weight; if the user input requirements do not mention the specified content, they are set to the default value when the large language model agent is constructed. In this embodiment, when the user enters "from 8 am to 6 pm today, energy costs need to be reduced first, the No. 2 and No. 3 cogeneration units and the energy storage system can be dynamically adjusted, and the remaining equipment still follows the day-ahead scheduling strategy", it is converted into an integrated energy system scheduling requirement, and an optimized scheduling scenario is constructed based on the integrated energy system scheduling requirement, which can be specifically expressed as:

[0062] Adjustable device collection as follows:

[0063]

[0064] in, They represent the cogeneration units numbered 2 and 3 and the energy storage equipment in the system topology library respectively.

[0065] Scheduling period as follows:

[0066]

[0067] in, To schedule the time, is the scheduling start time, , k is the number of scheduling, a total of N times; after parsing the user input, Set to 8:00, Set to 30 minutes and the total number of scheduling times to 20 times.

[0068] Constructing the scheduling target weight coefficient as follows:

[0069]

[0070] In step S4:

[0071] Based on the large language model agent, the historical price data and user input information are used to generate the carbon price forecast sequence and energy price forecast sequence within the scheduling period; the user input information includes real-time policy text and supply and demand situation information; if the user does not provide specified information, the forecast sequence is generated according to the default parameters of the large language model agent. Since the user input does not mention the real-time price policy and supply and demand situation information, the carbon price prediction sequence is constructed according to the default parameters of the large language model agent training. Energy price forecast series :

[0072]

[0073] in, is the time series trend analysis function, is the historical carbon price data, The carbon price weight coefficient for user input information, is the semantic influence quantification function, Enter information for the user.

[0074]

[0075] Where m is the energy type used by the equipment (including gas, steam, electricity), is the historical price data of energy of the corresponding type, The energy price weight coefficient corresponding to the type of user input information.

[0076] In step S5:

[0077] According to the optimized scheduling scenario described in step S3 and the carbon price and energy price predicted in step S4, a flexible optimized scheduling model including objective functions and constraints is established. Specifically, the flexible optimized scheduling model takes the weighted sum minimization of carbon cost and energy cost as the objective function, and takes the equipment output range constraint, equipment output ramp constraint, multi-energy flow balance constraint, and system topology constraint as the constraint conditions. By parsing the user input demand into the integrated energy system scheduling demand, an optimized scheduling scenario is constructed, and the carbon price and energy price prediction sequence are dynamically predicted. The established objective function and constraints can be flexibly changed to dynamically meet the actual scheduling needs. The objective function is specifically constructed as follows:

[0078]

[0079] in, is the objective function value, For device i Output during the period, ; Device i belongs to the adjustable device set ; is the carbon emission coefficient of equipment i.

[0080] The constraint parameter set is included in the domain knowledge and system data when the large language model is constructed, and the flexible optimization scheduling model can select a subset of the constraint parameter corresponding to the adjustable device set.

[0081] In step S6: solve the flexible optimization scheduling model described in step S5, and update the large language model agent in real time according to user feedback and scheduling results, and realize adaptive optimization of scheduling strategy based on the large language model agent. Figure 2 As shown, the specific steps are:

[0082] Solve the flexible optimization scheduling model to obtain adjustable equipment The optimal output at each dispatch time. Based on the user's further correction instructions for the obtained dispatch strategy and actual operation data, the large language model intelligent agent parses the correction instructions and extracts the integrated energy system dispatch requirements, dynamically updates the parameters of the flexible optimization dispatch model, and regenerates the dispatch strategy that meets the user's latest needs.

[0083] At the same time, the operation deviation data is obtained based on the correction instructions and actual operation data, and the user correction instructions and operation deviation data are input into the large language model intelligent agent. The price prediction logic of step S4 and the demand analysis rules of step S3 are updated through the incremental learning mechanism, and the multi-source heterogeneous knowledge base of the integrated energy system is optimized simultaneously to ensure that the subsequent scheduling strategy can adapt to external changes and ensure the effectiveness and robustness of the long-term scheduling effect.

[0084] Example 2

[0085] The present invention also provides a computer-readable storage medium having computer instructions stored thereon, wherein the computer instructions are used to enable a computer to execute the flexible optimization scheduling method for an integrated energy system based on a large language model.

[0086] It should be noted that the above is only part of the embodiments of the present invention. Equivalent changes made to the system described in the present invention are all included in the scope of protection of the present invention. Those skilled in the art of the present invention may make similar substitutions for the specific examples described, as long as they do not deviate from the structure of the present invention or exceed the scope defined by the claims, and all such substitutions are within the scope of protection of the present invention.

Claims

1. A flexible optimization scheduling method for an integrated energy system based on a large language model, characterized in that: The steps include: Step S1, constructing a multi-source heterogeneous knowledge base and rule logic coding system for the integrated energy system; Step S2: training a large language model agent with dynamic reasoning capabilities in the field of integrated energy systems based on the multi-source heterogeneous knowledge base and rule logic coding system of the integrated energy system; Step S3, using the large language model agent to parse the user input to obtain the integrated energy system scheduling requirements, wherein the integrated energy system scheduling requirements include adjustable equipment, scheduling period, and scheduling target weight; And build optimized scheduling scenarios based on the scheduling needs of the integrated energy system; Step S4, using the large language model agent to dynamically predict carbon prices and energy prices within the scheduling period; Step S5: establishing a flexible optimization scheduling model including an objective function and constraints based on the optimization scheduling scenario described in step S3 and the carbon price and energy price predicted in step S4; Step S6, solving the flexible optimization scheduling model described in step S5, and updating the large language model agent in real time according to user feedback and scheduling results, and realizing adaptive optimization of the scheduling strategy based on the large language model agent.

2. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 1 is characterized in that: The step S1 specifically includes the following steps: Step S11: Constructing a multi-source heterogeneous knowledge base for an integrated energy system. Specifically, the steps include integrating structured data and unstructured data, uniformly storing and associatively mapping the structured data and the unstructured data, and establishing a multi-source heterogeneous knowledge base for an integrated energy system. The multi-source heterogeneous knowledge base for an integrated energy system supports semantic retrieval. The structured data includes technical parameters of integrated energy system equipment, system topology, historical equipment and load data, and historical energy and carbon prices. The unstructured data includes trading mechanism rules and policy texts. Step S12, constructing a rule logic coding system; specifically: performing semantic parsing and entity relationship extraction on the unstructured data in the multi-source heterogeneous knowledge base of the integrated energy system, identifying the operating rules, constraints and policy clauses in the domain knowledge of the integrated energy system, and converting the operating rules, constraints and policy clauses into computer-executable formal logical expressions; binding the operating rules with the real-time state variables of the system through a dynamic mapping mechanism, generating a computable constraint set that adapts to different scheduling scenarios, and integrating the computable constraint set with the structured data in the multi-source heterogeneous knowledge base of the integrated energy system to obtain the rule logic coding system; the rule logic coding system supports multi-dimensional logical reasoning.

3. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 2 is characterized in that: In the step S2: Step S21, pre-training of basic capabilities; specifically, based on the pre-trained language model base, utilizing the multi-source heterogeneous knowledge base of the integrated energy system, establishing a mapping relationship between natural language instructions and adjustable equipment, scheduling periods, and scheduling target weights through supervised learning, thereby obtaining a large language model with the ability to semantically understand the terminology and basic rules in the field of the integrated energy system; and further strengthening the large language model's ability to semantically understand the terminology and basic rules in the field of the integrated energy system based on a rule-based logic coding system; Step S22: Enhanced training of complex reasoning capabilities. Specifically, through a multi-task joint training mechanism, the large language model is enabled to simultaneously learn semantic parsing, temporal trend reasoning, and dynamic constraint generation capabilities. At the same time, the large language model derives mathematical constraints that match the current scheduling scenario based on real-time device status, user demand priorities, and a rule-based logic coding system, and calls the computable constraint set to verify logical consistency, thereby improving the complex reasoning capabilities of the large language model. Step S23, incremental learning interface deployment; Specifically, the large language model dynamically accesses equipment change parameters, policy change texts and user feedback information, and realizes the continuous iteration of the cognitive ability of the large language model through efficient parameter fine-tuning methods, and synchronously updates the rules and associations of the constraints in the rule logic coding system to ensure the long-term adaptability of the large language model's reasoning ability and the dynamic evolution of the system, thereby obtaining a large language model intelligent entity with dynamic reasoning capabilities in the field of integrated energy systems.

4. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 1 is characterized in that: The step S3 is specifically as follows: Based on the natural language understanding capabilities of the large language model, the text scheduling requirements entered by the user are converted into integrated energy system scheduling requirements. If the user input requirements do not mention the specified content, they are set as the default value when the large language model agent is constructed; An optimized scheduling scenario is constructed based on the scheduling requirements of the integrated energy system, which is specifically expressed as follows: Adjustable device collection for: ； in, A collection of all devices in the system. Representation device At the real-time state at time t, if the state is 1, it is adjustable; The scheduling period T is: ； in, To schedule the time, is the scheduling start time, , k is the number of scheduling times, a total of N scheduling times; Scheduling target weight parameter for: ； in, is the energy cost weight, is the carbon emission cost weight.

5. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 4 is characterized in that: The step S4 is specifically as follows: Based on the large language model intelligent agent, historical price data and user input information are used to generate carbon price forecast sequences and energy price forecast sequences within the scheduling period; user input information includes real-time policy text and supply and demand situation information; if the user does not provide specified information, the forecast sequence is generated according to the default parameters of the large language model intelligent agent.

6. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 5 is characterized in that: The carbon price prediction series for: ； in, is the time series trend analysis function, is the historical carbon price data, The carbon price weight coefficient for user input information, is the semantic influence quantification function, Enter information for the user; The energy price forecast series for: ； Where m is the type of energy used by the equipment, is the historical price data of energy of the corresponding type, The energy price weight coefficient for the corresponding type in the user input information.

7. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 6 is characterized in that: In the step S5, the objective function is to minimize the weighted sum of carbon cost and energy cost, and the constraints are equipment output range constraint, equipment output ramp constraint, multi-energy flow balance constraint, and system topology constraint.

8. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 7 is characterized in that: The objective function is: ； in, is the objective function value, For device i Output of the time period, device i belongs to the set of adjustable devices , is the carbon emission coefficient of equipment i.

9. The method for flexible optimization and scheduling of an integrated energy system based on a large language model according to claim 1, characterized in that: In the step S6: The large language model agent is updated in real time based on user feedback and scheduling results. The specific method is as follows: Based on the user's further correction instructions for the obtained scheduling strategy and actual operation data, the large language model intelligent agent parses the correction instructions and extracts the scheduling requirements of the integrated energy system, dynamically updates and flexibly optimizes the scheduling model parameters, and regenerates the scheduling strategy that meets the user's latest needs; at the same time, based on the correction instructions and actual operation data, the operation deviation data is obtained, and the correction instructions and operation deviation data are input into the large language model intelligent agent. The price prediction logic of step S4 and the scheduling requirement analysis rules of step S3 are updated through the incremental learning mechanism, and the multi-source heterogeneous knowledge base of the integrated energy system is simultaneously optimized to ensure that the subsequent scheduling strategy can adapt to external changes.

10. A computer-readable storage medium having computer instructions stored thereon, characterized in that: The computer instructions are used to enable a computer to execute the steps of the method according to any one of claims 1 to 9.

Images