User demand driven distributed energy supply system capacity configuration method based on large language model

Abstract

The invention discloses a user demand-driven distributed energy supply system capacity configuration method based on a large language model. The method comprises the following steps: collecting user demand information; performing semantic analysis and understanding on the acquired user demand information by utilizing a large language model, and outputting structured and quantitative user demand description; parameterizing the structured and quantized user demand description to obtain a mathematical model for capacity configuration optimization; and solving a mathematical model for capacity configuration optimization by adopting a numerical optimization algorithm, generating a plurality of optimization schemes, and performing performance evaluation and visual display to assist a user in selection. According to the method, the accuracy and the intelligent level of user demand understanding are improved through the large language model, automatic processing and personalized capacity configuration of unstructured demands are achieved, the capacity configuration efficiency and the user satisfaction degree of the distributed energy supply system are remarkably improved, and the system performance is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of energy engineering, specifically relates to the optimization of distributed energy system capacity configuration, and particularly relates to a method for analyzing user requirements using a large language model and realizing the capacity configuration of a distributed energy supply system in combination with numerical optimization methods. Background Art

[0002] With the urgent need for global energy structure transformation and sustainable development, the distributed energy supply system (DES), as an efficient, clean, and flexible energy supply mode, has been rapidly developed and widely applied worldwide. The distributed energy supply system is usually deployed on the user side or near the user, and uses renewable energy (such as solar energy, wind energy) and clean energy (such as natural gas) to realize the in-situ production and nearby utilization of various energy forms such as electricity, heat, and cold, with multiple advantages such as improving energy utilization efficiency, reducing power transmission and distribution losses, enhancing energy supply reliability, and promoting the consumption of renewable energy.

[0003] The core limitations of existing distributed energy supply system deployment methods are mainly reflected in the following aspects: First, in terms of understanding user requirements, existing methods are difficult to accurately capture the true intentions and personalized preferences of users, resulting in a low matching degree between the final configuration scheme and user requirements, and unable to achieve personalized configuration driven by user needs in the true sense. Second, for the unstructured requirements that users usually express in natural language, existing methods are difficult to directly and effectively process, and often require manual conversion and interpretation, which is not only inefficient but also error-prone. Third, the level of intelligence and efficiency in the capacity configuration process is low, mainly relying on manual experience and manual calculations, and it is difficult to cope with the increasingly complex system design requirements. Finally, in the face of large-scale complex systems and the inherent uncertainties of the operating environment, existing methods are unable to handle model construction, algorithm solving, and ensuring the robustness of the optimization scheme. Summary of the Invention

[0004] To break through the bottleneck of the existing technology, the present invention provides a method for configuring the capacity of a distributed energy supply system driven by user requirements based on a large language model. The core idea of this method is to use the powerful natural language processing and semantic understanding capabilities of the large language model to deeply analyze and understand the capacity configuration requirements of the distributed energy supply system expressed by users in unstructured forms such as natural language, and convert the understanding results into structured and quantified requirement descriptions, thereby driving the subsequent generation and optimization process of the capacity configuration scheme.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for configuring the capacity of a distributed energy supply system driven by user needs based on a large language model, comprising the following steps:

[0007] Step S1, collect user demand information, where the user demand information includes at least one or a combination of natural language text, structured questionnaires, and historical energy consumption data;

[0008] Step S2, use a large language model to perform semantic analysis and understanding on the collected user demand information, and output a structured and quantified user demand description, where the structured and quantified user demand description includes requirements, demand priorities, and constraint conditions;

[0009] Step S3, parameterize the structured and quantified user demand description to obtain a mathematical model for capacity configuration optimization, where the mathematical model for capacity configuration optimization includes an objective function and constraint conditions, and the objective function and constraint conditions are set based on the structured and quantified user demand description;

[0010] Step S4, based on the mathematical model for capacity configuration optimization, use a numerical optimization algorithm to generate at least one distributed energy supply system capacity configuration plan, and perform performance evaluation and visual display on the distributed energy supply system capacity configuration plan.

[0011] Furthermore, in the above-mentioned step S1:

[0012] In step S11, the method for obtaining natural language text is that the user inputs through a text box, voice input conversion, etc., and the user can freely describe their expectations for the distributed energy supply system, such as expected performance goals (economy, reliability, environmental protection), preferred energy types, specific technical requirements or limiting conditions, etc.

[0013] In step S12, the method for obtaining a structured questionnaire is that the user fills in a pre-set structured questionnaire, and the structured questionnaire specifically includes multiple pre-defined questions covering key aspects of the capacity configuration of the distributed energy supply system. These questions guide the user to provide structured and directly quantifiable information, such as the expected return on investment, acceptable power outage duration, expected value of the proportion of renewable energy, etc., to ensure that the parameters and constraint conditions required for capacity configuration optimization are collected.

[0014] In step S13, the method for obtaining historical energy consumption data is that the user imports historical energy consumption data, such as CSV files exported by smart meters, scanned copies of electricity bills, etc. Historical energy consumption data reflects the user's actual energy consumption pattern, such as the time distribution of electricity consumption, peak-valley characteristics, seasonal changes, etc., and is used to provide a demand-side basis for capacity configuration optimization.

[0015] Furthermore, in the above-mentioned step S2:

[0016] Step S21, intention recognition and requirement classification: Utilize the intention recognition ability of the large language model to recognize the true intention contained in the user requirement information, so as to obtain the core intention of the user, such as "seeking the most economical solution" or "pursuing the highest power supply reliability". At the same time, classify the user requirement information according to predefined categories, such as classifying the requirements into categories like "economic requirements", "reliability requirements", "environmental preferences", "functional requirements", "constraint conditions", etc. The classification process can adopt a multi-label classification model, allowing a user requirement to belong to multiple categories simultaneously.

[0017] Step S22, extraction and quantification of key features of requirements: Utilize the information extraction and knowledge reasoning abilities of the large language model to extract key feature words, phrases or sentences for each requirement category from the classified user requirement information. And further transform these qualitative feature descriptions into quantitative indicators. For example, extract the feature of "reducing investment cost" from "hoping to minimize the investment cost" and quantify it as "upper limit of initial investment (ten thousand yuan)".

[0018] Step S23, requirement priority ranking: According to the preferences and emphasis degrees in the user requirement information, utilize the large language model to judge the relative importance of different requirements and conduct priority ranking to provide a basis for subsequent multi-objective optimization.

[0019] Step S24, identification and extraction of constraint conditions: Utilize the large language model to identify and extract explicit or implicit constraint conditions from the user requirement information, where the constraint conditions are technical limitations, economic budgets, and environmental requirements.

[0020] Furthermore, in step S3:

[0021] Step S31, requirement parameter mapping: Map the requirements, quantitative indicators of requirements, priority ranking, and constraint condition information extracted in step S2 into parameters available in the mathematical model for capacity configuration optimization.

[0022] Step S32, construct the objective function of the mathematical model for capacity configuration optimization according to the parameters obtained in step S31, where the objective function is a single-objective optimization or multi-objective optimization function.

[0023] Step S33, constraint condition setting: Set the constraint conditions of the capacity configuration optimization model according to the parameters related to the constraint conditions obtained in step S31 and the actual constraints of the distributed energy supply system operation, where the constraint conditions include technical constraints, economic constraints, and environmental constraints.

[0024] Furthermore, in step S4:

[0025] Step S41: Generation of the capacity configuration plan for the distributed energy supply system. Select a suitable numerical optimization algorithm to solve the mathematical model for capacity configuration optimization, and generate one or more candidate capacity configuration plans for the distributed energy supply system.

[0026] Step S42: Performance evaluation of the plan and calculation of indicators. Conduct performance evaluation on the generated candidate capacity configuration plans, and calculate performance indicators such as economy, reliability, and environmental friendliness of the plans.

[0027] Step S43: Ranking and selection assistance for multiple plans. If multiple candidate plans are generated, rank or classify the multiple plans according to the demand priorities and performance indicators set by the user to assist the user in making a choice.

[0028] Step S44: Visual display of the evaluation results. Visually display the performance evaluation results of the generated capacity configuration plan and the comparison information of multiple plans.

[0029] The present invention innovatively proposes a method for driving the capacity configuration of a DES based on a large language model according to user needs. Its core innovation lies in that for the first time, the large language model (LLM) technology, which has achieved great breakthroughs in the field of natural language processing in recent years, is creatively introduced into the field of DES capacity configuration. The present invention makes full use of the powerful semantic understanding and knowledge reasoning capabilities of the LLM to achieve the following key technological breakthroughs: innovatively introducing the large language model, which can deeply analyze the needs expressed by users in natural language, accurately identify user intentions and preferences, significantly improving the accuracy and depth of user need understanding, and effectively overcoming the deficiencies of traditional methods in user need understanding; it can directly process unstructured user needs such as natural language without manual intervention, realizing the automation and intelligence of need analysis, greatly improving efficiency and reducing labor costs; the present invention is driven by the user needs intelligently understood by the large language model, automatically generating and recommending personalized capacity configuration plans, realizing the intelligent transformation from user needs to plans, significantly improving the user satisfaction of the plans and the system performance; furthermore, the present invention deeply integrates the large language model technology with numerical optimization algorithms, realizing the intelligence and automation of the entire process of capacity configuration, greatly improving the efficiency and intelligence level of capacity configuration, and effectively reducing the dependence on professional personnel.

[0030] Compared with the prior art, the beneficial effects of the present invention are:

[0031] The method for driving the capacity configuration of a distributed energy supply system based on a large language model according to user needs provided by the present invention has the following remarkable beneficial effects compared with the prior art:

[0032] Utilizing the powerful natural language processing and semantic understanding capabilities of large language models, it is possible to more accurately and deeply understand the capacity configuration requirements of distributed energy supply systems expressed by users in unstructured forms such as natural language, overcoming the bottleneck problem of insufficient understanding of user requirements in traditional methods, significantly improving the accuracy and depth of user requirement understanding, and laying a solid requirement foundation for the subsequent generation and optimization of capacity configuration plans.

[0033] User demand-driven customized capacity configuration: The method of the present invention is driven by the user requirements intelligently understood by the large language model, guiding the subsequent process of generating and optimizing the capacity configuration plan. It can automatically generate and recommend customized capacity configuration plans according to the personalized requirements of different users, realizing the intelligent transformation from user requirements to configuration plans, significantly improving the matching degree between the configuration plan and user requirements and user satisfaction, and enhancing the system performance.

[0034] Intelligent and efficient capacity configuration process: The method of the present invention combines large language model technology with numerical optimization algorithms, realizing the intelligence and automation of the entire capacity configuration process including user requirement analysis, plan generation, plan evaluation, and plan display, greatly improving the efficiency and intelligence level of capacity configuration, shortening the capacity configuration cycle, reducing the dependence on professional personnel, and providing strong support for the rapid popularization and application of distributed energy supply systems.

[0035] The proposal of the technical solution of the present invention truly breaks through the bottlenecks of traditional DES capacity configuration methods in user requirement understanding and intelligence, provides a new technical approach for realizing more intelligent, efficient, and personalized DES capacity configuration, has significant innovation and practical value, and is expected to promote the wider application of DES technology. Brief Description of the Drawings

[0036] The present invention will be further described below in conjunction with the drawings and embodiments.

[0037] Figure 1 It is the main step diagram of the method of the present invention.

[0038] Figure 2 It is the schematic diagram of the structured questionnaire. Detailed Embodiment

[0039] The following combines the Figure 1 The flowchart of the capacity configuration method for distributed energy supply systems driven by user requirements based on large language models shown below, and a specific embodiment of the capacity configuration of a household photovoltaic energy storage system, to describe the technical solution of the present invention in detail.

[0040] Example: Capacity Configuration of Household Photovoltaic Energy Storage System

[0041] Suppose the user is a family living in the suburbs of a certain city, and they hope to install a photovoltaic energy storage system on the roof of their house to reduce the household electricity cost and improve the electricity reliability. The user can design a plan through the distributed energy supply system capacity configuration method of the present invention. The specific configuration method includes the following steps:

[0042] Step S1, collect user demand information:

[0043] Step S11, natural language text input. The user inputs a natural language description: "I hope to install photovoltaic panels on the roof of my house. The main purpose is to save electricity bills, and it is best to use as much electricity generated by solar energy as possible. In addition, if there is a power outage, I hope there can be a backup power supply at home to ensure that the refrigerator and lighting can be used normally for several hours. In terms of budget, I hope the total investment can be controlled within 50,000 yuan.

[0044] Step S12, structured questionnaire: The user fills in a structured questionnaire. The structured questionnaire specifically includes multiple pre-defined questions covering key aspects of the distributed energy supply system capacity configuration. These questions guide the user to provide structured and directly quantifiable information to ensure that the parameters and constraints required for capacity configuration optimization are collected. Some key questions and examples of user filling are Figure 2 as shown.

[0045] Step S13, import historical energy consumption data. The user uploads a CSV-format electricity consumption data file exported from the electricity meter in the past year, and the file contains the electricity consumption records per hour.

[0046] In addition, the system can also collect information such as environment, policies, and electricity prices related to capacity configuration. For example: The user authorizes the system to obtain geographical location information. The system automatically obtains the location of the user's city and suburbs, and obtains meteorological data such as the average sunshine duration and solar radiation intensity in this area in recent years from the meteorological data API interface; obtains the local residential electricity price of about 0.6 yuan per kilowatt-hour from the power grid information database; obtains the on-grid electricity price subsidy policy for residential-side photovoltaic power generation in this area as 0.08 yuan per kilowatt-hour from the policy information database.

[0047] Step S2, use a large language model to perform semantic analysis and understanding on the collected user demand information, and output a structured and quantified user demand description. The structured and quantified user demand description includes requirements, requirement priorities, and constraints:

[0048] Step S21, intention recognition and requirement classification: Use a large language model to analyze the user demand text and the structured questionnaire, and recognize that the user's intention is "hoping to configure a family photovoltaic energy storage system with good economy and certain reliability". At the same time, classify the user's demand information according to pre-defined categories. For example, classify the requirements into:

[0049] Economic requirements: Save electricity costs, control the investment budget within 50,000 yuan, and expect an investment payback period of 5 - 8 years.

[0050] Reliability requirements: Hope to avoid power outages as much as possible. When there is a power outage, the backup power supply can maintain the power supply for the refrigerator and lighting for several hours.

[0051] Environmental protection preference: Hope to use as much electricity generated by solar energy as possible and tend to use clean energy.

[0052] Constraint conditions: The total investment budget is within 50,000 yuan, and the roof can be installed with photovoltaic panels.

[0053] Step S22, extraction and quantification of key features of requirements: Use the large language model to extract key features (such as key feature words, phrases or sentences) from the classified requirement information and quantify them:

[0054] Economic features: "Save electricity costs" is quantified as "Expected annual average electricity cost savings ratio > 20%" (assuming estimated based on historical data, electricity price and potential photovoltaic output); "Investment budget within 50,000 yuan" is quantified as "Initial investment upper limit 50,000 yuan"; "Investment payback period of 5 - 8 years" is quantified as "Investment payback period upper limit 8 years".

[0055] Reliability features: "Avoid power outages as much as possible" is quantified as "System availability > 99.9%"; "Backup power supply maintains the refrigerator and lighting for several hours" is quantified as "Energy storage system backup duration > 4 hours" (assuming estimated based on user checkboxes and equipment power).

[0056] Environmental protection preference features: "Use as much solar energy as possible" is quantified as "Proportion of photovoltaic power generation in total electricity consumption > 50%".

[0057] Step S23, requirement priority ranking: The large language model analyzes the preferences and emphasis levels in the user requirement information, judges the relative importance of the user's economic, reliability, environmental protection and other requirements, and ranks them as: economy > reliability > environmental protection. And according to the ranking results, initially set the weights: economic weight 0.5, reliability weight 0.3, environmental protection weight 0.2 (the weight values can be adjusted later).

[0058] Step S24, identification and extraction of constraint conditions: Use the large language model to identify and extract explicit or implicit constraint conditions from the user requirement information. The constraint conditions are technical limitations, economic budgets, and environmental protection requirements.

[0059] Extract the economic budget constraint from natural language and structured questionnaires: The total investment budget is within 50,000 yuan C max = (50,000 yuan).

[0060] Extracting technical limitations from structured questionnaires: The installable area on the roof is approximately 30 square meters (which can be converted into the maximum photovoltaic capacity constraint according to the component power density).

[0061] Identifying potential constraints or preferences related to environmental protection requirements from user intentions: Try to use more solar energy (converted into the requirement for the utilization rate of renewable energy).

[0062] Step S3: Parametrize the structured and quantified user requirement descriptions to obtain a mathematical model for capacity configuration optimization. The mathematical model for capacity configuration optimization includes an objective function and constraint conditions, and the objective function and constraint conditions are set based on the structured and quantified user requirement descriptions:

[0063] Step S31: Requirement parameter mapping: Map the requirements, quantification metrics of the requirements, priority rankings, and constraint condition information extracted in step S2 into parameters that can be used in the mathematical model for capacity configuration optimization. These parameters will be directly used to construct the objective function and constraint conditions of the model:

[0064] Initial investment upper limit parameter: C max = 50000 (yuan)

[0065] Investment payback period upper limit parameter: PBP = 8 (years)

[0066] Expected annual electricity cost savings ratio parameter: SR min = 0.2

[0067] System availability lower limit parameter: Avail min = 0.999

[0068] Energy storage backup duration lower limit parameter: T min = 4 hours)

[0069] Renewable energy power generation ratio lower limit parameter: RR min = 0.5

[0070] Objective weight parameter set based on priority: W economic ,W reliability ,W environmental (For example, set as W according to the priority of economy > reliability > environmental protection economic = 0.5,W reliability =

[0071] 0.3,W environmental = 0.2)

[0072] Roof area limit parameter: Roof max = 30 (square meters) converted into the maximum photovoltaic capacity limit parameter PV_{max}

[0073] Step S32, objective function construction: According to the parameters obtained in step S31, construct the objective function of the mathematical model for optimizing the capacity configuration. Considering the user's requirements for economy, reliability, and environmental protection, construct a multi-objective optimization function. The weighted summation method or the Pareto optimization method can be used. Taking the weighted summation method as an example, the objective function is set to minimize the comprehensive cost or maximize the comprehensive benefit, and indicators such as economy (e.g., life cycle cost), reliability (e.g., availability or power outage loss), and environmental protection (e.g., renewable energy utilization rate) are used as optimization objectives.

[0074] minimizeF=W economic ·f economic -W reliability ·f reliability -W environmental ·f environmental

[0075] where F is the comprehensive cost, the life cycle cost (f economic ) is the main objective, the system availability (f reliability ) and the renewable energy utilization rate (f environmental ) are the secondary objectives, and the weight coefficients are determined by step S23.

[0076] Step S33, constraint condition setting: According to the parameters related to the constraint conditions obtained in step S31 and the actual constraints of the distributed energy supply system operation, set the constraint conditions of the capacity configuration optimization model, including:

[0077] Technical constraints: upper and lower limits of the capacities of devices such as photovoltaic modules, energy storage batteries, and inverters, efficiency characteristics, operation constraints, etc. Energy balance constraint (photovoltaic power generation + grid power purchase = load power consumption + energy storage charging + energy loss). Energy storage charge and discharge power constraint, charge and discharge depth constraint, etc. Roof installable area constraint (assuming the user's roof installable area is 30 square meters, converted to the maximum installable photovoltaic capacity according to the power density of the photovoltaic module). System availability constraint (system availability ≥ Avail min = 0.999), energy storage backup duration constraint (energy storage backup duration ≥ T min = 4 hours).

[0078] Economic constraints: total investment cost constraint (total investment cost ≤ C max = 50,000 yuan), payback period constraint (payback period ≤ PBP = 8 years).

[0079] Environmental protection constraints: renewable energy utilization rate constraint (proportion of renewable energy power generation ≥ RR min = 0.5).

[0080] Step S4. Based on the mathematical model for capacity configuration optimization, use a numerical optimization algorithm to generate at least one capacity configuration plan for the distributed energy supply system, and perform performance evaluation and visual display on the capacity configuration plan of the distributed energy supply system:

[0081] Step S41. Generation of the capacity configuration plan for the distributed energy supply system: Select a suitable numerical optimization algorithm, such as simulation-based optimization algorithms (such as genetic algorithms, particle swarm algorithms), mixed integer linear programming (MILP), or enumeration method (for small-scale systems). In this embodiment, the genetic algorithm is selected to solve the mathematical model constructed in Step S3. The decision variables of the algorithm are the capacities of the key devices of the distributed energy supply system, such as the total capacity of photovoltaic modules (kWp), the total capacity of energy storage batteries (kWh), and the capacity of the inverter (kW). The algorithm searches within the discrete set or continuous range of device capacities to solve for the device capacity combination that optimizes the objective function constructed in Step S32 (or reaches the Pareto optimal front) and satisfies all the constraint conditions in Step S33, thereby generating one or more candidate capacity configuration plans for the distributed energy supply system. For example, after the genetic algorithm iteratively runs, it outputs multiple plans that meet the constraint conditions, and each plan includes a specific set of photovoltaic capacity, energy storage capacity, and inverter capacity.

[0082] Step S42. Scheme performance evaluation and index calculation: Conduct a detailed performance evaluation on each candidate capacity configuration plan generated in Step S41. The evaluation process usually adopts the simulation method, combined with the historical energy consumption data provided by the user (Step S13), meteorological data (Step S14), electricity price, and policy information (Step S14), to simulate the operation of the system on a long time scale (such as one year, hour by hour). According to the simulation results, calculate the performance indicators of the plan:

[0083] Economic indicators: Initial investment cost, operation and maintenance cost, levelized cost of electricity (LCOE), payback period, and annual average electricity cost savings.

[0084] Reliability indicators: System availability, load satisfaction rate, power outage loss cost, and energy storage backup duration.

[0085] Environmental protection indicators: Renewable energy utilization rate, CO2 emission reduction, and energy self-sufficiency rate.

[0086] For example, for a generated plan (such as 5 kW photovoltaic + 8 kWh energy storage), calculate its operation data under the user's annual electricity consumption curve and local meteorological conditions through hourly simulation, and calculate that the initial investment cost of this plan is 48,000 yuan, the levelized cost of electricity is 0.4 yuan / kWh, the payback period is 6.5 years, the system availability is 99.95%, the energy storage backup duration is 5 hours, and the renewable energy utilization rate is 60%.

[0087] Step S43, Multi-Scheme Sorting and Selection Assistance: If multiple candidate schemes are generated in Step S41 (such as multiple points on the Pareto optimal frontier), the system sorts or classifies the multiple candidate schemes according to the user-set requirement priorities (Step S23) and the performance metrics calculated in Step S42 to assist the user in making a selection. For example, the schemes can be scored according to the priority order, or the schemes can be visualized in a two-dimensional / three-dimensional coordinate system (such as economy vs reliability) for the user to intuitively compare. According to the priority of economy > reliability > environmental protection in this example, the system can first display the schemes with low initial investment or short payback period, and then further sort them by availability rate among these schemes. Finally, several recommended schemes (such as the Top3 schemes) are selected and presented to the user.

[0088] Step S44, Visual Display of Evaluation Results: The performance evaluation results of the capacity configuration scheme calculated in Step S42 (for the multi-scheme case, also including the comparison information in Step S43) are visually displayed so that the user can intuitively understand the advantages and disadvantages of different schemes. The visualization forms can include:

[0089] Scheme Summary Table: List the key equipment capacity combinations and core performance metric values of each recommended scheme.

[0090] Performance Metric Comparison Chart: Use bar charts, radar charts or line charts to compare the differences in indicators such as initial investment, payback period, availability rate, and renewable energy utilization rate among different schemes.

[0091] System Operation Simulation Curve: Plot the power or energy curves of hourly photovoltaic power generation, load power consumption, energy storage charging and discharging, and grid interaction on a typical day or throughout the year to show the operation characteristics of the scheme.

[0092] Economic Benefit Analysis Chart: Show the curve of cumulative electricity cost savings over time, the schematic diagram of the investment payback period, etc.

Claims

1. A method for configuring the capacity of a distributed energy supply system driven by user needs based on a large language model, characterized in that, It includes the following steps: Step S1, collect user requirement information, where the user requirement information includes at least one or a combination of natural language text, structured questionnaire, and historical energy consumption data; Step S2, use a large language model to perform semantic analysis and understanding on the collected user requirement information, and output a structured and quantified user requirement description, where the structured and quantified user requirement description includes requirements, requirement priorities, and constraint conditions; Step S3, parameterize the structured and quantified user requirement description to obtain a mathematical model for capacity configuration optimization, where the mathematical model for capacity configuration optimization includes an objective function and constraint conditions, and the objective function and constraint conditions are set based on the structured and quantified user requirement description; Step S4, based on the mathematical model for capacity configuration optimization, use a numerical optimization algorithm to generate at least one distributed energy supply system capacity configuration plan, and perform performance evaluation and visual display on the distributed energy supply system capacity configuration plan.

2. A method for configuring the capacity of a distributed energy supply system driven by user needs based on a large language model according to claim 1, characterized in that, In the above-mentioned step S1: The user inputs natural language text through a text box or voice input conversion method, and the natural language text is used to describe the expected performance goals, preferred energy types, technical requirements, or limiting conditions of the user for the distributed energy supply system; The structured questionnaire contains predefined questions to guide the user to provide structured and directly quantifiable information for collecting the parameters and constraint conditions required for capacity configuration optimization; The historical energy consumption data is used to reflect the user's actual energy consumption pattern and provide a demand-side basis for capacity configuration optimization.

3. A method for configuring the capacity of a distributed energy supply system driven by user needs based on a large language model according to claim 2, characterized in that In the above-mentioned step S2: Step S21, intention recognition and requirement classification, use the intention recognition ability of the large language model to recognize the true intention contained in the user requirement information, and classify the user requirement information according to predefined categories; Step S22, extraction and quantification of key features of requirements, use the information extraction and knowledge reasoning abilities of the large language model to extract key feature words, phrases, or sentences for each requirement category from the classified user requirement information, and convert the qualitative feature descriptions into quantitative indicators; Step S23, determination of requirement priority ranking, judge the relative importance of different requirements according to the preferences and emphasis degrees in the user requirement information using the large language model, and perform priority ranking; Step S24, identification and extraction of constraint conditions, use the large language model to identify and extract explicit or implicit constraint conditions from the user requirement information, where the constraint conditions include technical limitations, economic budgets, and environmental protection requirements.

4. A method for configuring the capacity of a distributed energy supply system driven by user requirements based on a large language model according to claim 3, characterized in that, In the above-mentioned step S3: Step S31, requirement parameter mapping, map the requirement, quantitative indicators of the requirement, priority ranking, and constraint condition information extracted in step S2 into parameters available in the mathematical model for capacity configuration optimization; Step S32, construction of the objective function, construct the objective function of the mathematical model for capacity configuration optimization according to the parameters obtained in step S31, and the objective function is a single-objective optimization or multi-objective optimization function; Step S33, Constraint setting: According to the parameters related to the constraints obtained in Step S31 and the actual constraints of the distributed energy supply system operation, set the constraint conditions of the capacity configuration optimization model, where the constraint conditions include technical constraints, economic constraints, and environmental protection constraints.

5. A method for capacity configuration of a distributed energy supply system driven by user requirements based on a large language model according to claim 4, characterized in that In the said Step S4: Step S41, Generation of distributed energy supply system capacity configuration plans: Use numerical optimization algorithms to solve the mathematical model for capacity configuration optimization, and generate one or more candidate distributed energy supply system capacity configuration plans; Step S42, Performance evaluation and index calculation of distributed energy supply system capacity configuration plans: Conduct performance evaluations on the generated candidate capacity configuration plans, and calculate the economic, reliability, and environmental protection indicators of the plans; Step S43, Multi-plan ranking and selection assistance: If multiple plans are generated, rank or classify the multiple plans according to the demand priorities and performance indicators set by the user to assist the user in making a selection; Step S44, Visual display of evaluation results: Visually display the performance evaluation results of the generated capacity configuration plans and the multi-plan comparison information.

Images