Micro-grid configuration system with optimal power consumption cost under AI prediction-based electricity market

Abstract

This invention discloses an AI-based microgrid configuration system for optimizing electricity costs in a power market, belonging to the field of power system and renewable energy configuration technology. The system collects various types of raw data through a data acquisition module and preprocesses them to obtain standard data. An AI prediction module uses this standard data to predict electricity prices and load time series and outputs load fluctuation warning data. An electricity consumption diagnosis module combines raw data and AI prediction data to perform four-dimensional dynamic diagnosis, outputting quantitative diagnostic results and data on existing electricity consumption pain points. An optimal configuration module uses the standard data and data output by the AI ​​prediction module, relying on multi-layered constraints to solve for the optimal installed capacity layout and collaborative operation strategy for solar power, energy storage, and wind power. A results output module visualizes various data and generates standardized reports. This invention solves the problems of traditional solutions being unable to adapt to dynamic electricity prices in the power market, having low prediction accuracy, and poor targeting of configuration schemes, thereby reducing user electricity costs and adapting to renewable energy collaborative optimization configuration scenarios.

Description

Technical Field

[0001] This invention belongs to the field of power system diagnosis and new energy configuration technology, and in particular relates to a microgrid configuration system with optimal electricity cost in the power market based on AI prediction. Background Technology

[0002] With the deepening of my country's power system reform, some provinces have gradually abolished the administrative division of peak and off-peak electricity periods, and most industrial and commercial electricity users have been included in the electricity market for market-based electricity purchase. Electricity prices are no longer fixed administrative prices, but fluctuate dynamically with market supply and demand, exhibiting characteristics such as randomness, volatility, and cyclicality. At the same time, new energy technologies such as distributed photovoltaic, wind power, and energy storage are developing rapidly, becoming the core path for industrial and commercial users to reduce electricity costs and achieve a zero-carbon transition.

[0003] Currently, existing electricity consumption diagnostic systems mainly rely on historical fixed electricity prices and measured electricity consumption data to conduct energy consumption analysis and recommend renewable energy configurations, which has the following core shortcomings: 1. Unable to adapt to dynamic electricity market pricing scenarios: The existing system does not consider the dynamic fluctuation characteristics of electricity prices in the electricity market environment, and still uses fixed peak-valley electricity prices or administrative pricing for cost calculation and configuration optimization. This results in the recommended photovoltaic, wind power, and energy storage configuration schemes being out of touch with the actual market-based electricity purchase scenario, and failing to achieve the goal of the lowest electricity cost. 2. Lack of accurate forecasting capabilities: Existing systems rely heavily on historical data for static analysis and do not incorporate AI forecasting technology. This makes it impossible to accurately predict future electricity load and electricity market prices, resulting in a lack of foresight in renewable energy allocation schemes. This can easily lead to over- or under-allocation, affecting revenue and cost control. 3. The configuration optimization objectives are singular and disconnected: The existing system's new energy configuration is mostly aimed at maximizing power generation and carbon emission reduction, without taking the lowest electricity cost as the core constraint. Furthermore, it has not achieved coordinated optimization of photovoltaic, wind power and energy storage configurations, ignoring the complementary characteristics of their outputs (the diurnal periodicity of photovoltaics, the intermittency of wind power, and the flexible dispatchability of energy storage), and thus cannot take into account cost, benefits and low-carbon objectives. 4. Insufficient coordination between diagnosis and configuration: The existing system's power consumption diagnosis, prediction, and new energy configuration are independent modules, and the data cannot be interconnected. The diagnosis results cannot directly guide the configuration optimization, and the configuration scheme cannot be adapted to the optimization needs of power consumption diagnosis, forming data silos and affecting the overall usability of the system.

[0004] Furthermore, in existing technologies, AI-based electricity price forecasting and load forecasting are mostly independent technologies that are not deeply integrated with electricity consumption diagnosis and new energy configuration. They are also not customized for the electricity consumption characteristics of industrial and commercial users (stable production load, cost sensitivity, and prominent zero-carbon demand), and therefore cannot meet the dual needs of industrial and commercial users for "cost reduction and efficiency improvement + zero-carbon transformation" in the electricity market environment.

[0005] The purpose of this invention is to provide a microgrid configuration system based on AI prediction for optimal electricity costs in the electricity market, in order to solve the problems existing in the prior art. Summary of the Invention

[0006] To achieve the above objectives, the present invention provides the following solution: The present invention provides a microgrid configuration system for optimal electricity cost under an AI-predicted electricity market, the system comprising: The data acquisition module is used to collect raw data and perform preprocessing such preprocessing, including cleaning, verification, and normalization, as well as removing invalid and abnormal data to obtain preprocessed standard data. The raw data includes basic raw data on electricity consumption of industrial and commercial users, dynamic raw data on the electricity market, and auxiliary support raw data for new energy construction. The AI ​​prediction module is used to build AI electricity price prediction models and AI load prediction models. Based on the preprocessed standard data, it completes dynamic predictions of electricity market prices and user electricity loads at multiple time scales in the future, and obtains time-of-use electricity price prediction data, electricity load prediction data, and electricity price and load fluctuation early warning data. The electricity consumption diagnosis module is used to perform dynamic electricity consumption diagnosis based on the original electricity consumption data of industrial and commercial users, the time-of-use electricity price prediction data, the electricity load prediction data, and the electricity price and load fluctuation early warning data. It can also output quantitative diagnosis results and existing electricity consumption pain point data. The optimal configuration module is used to output the optimal installed capacity and layout scheme of photovoltaic, wind power and energy storage integration based on the preprocessed standard data, the time-of-use electricity price prediction data, the electricity load prediction data, the electricity price and load fluctuation early warning data, the quantitative diagnosis results and the existing electricity consumption pain point data, and at the same time generate a photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations. The results output module is used to visualize the time-of-use electricity price forecast data, the electricity load forecast data, the electricity price and load fluctuation early warning data, the quantitative diagnostic results, the existing electricity consumption pain point data, the optimal installed capacity, the layout scheme, and the photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations. It also generates an exportable standardized configuration diagnostic report and supports manual parameter adjustment and scheme recalculation.

[0007] Optionally, the data acquisition module includes: The industrial and commercial user electricity consumption basic raw data acquisition submodule is used to collect the industrial and commercial user electricity consumption basic raw data, which includes active power, reactive power, voltage, current, time-of-use electricity consumption, power factor, and load factor. The electricity market dynamic raw data acquisition submodule is used to collect the electricity market dynamic raw data, which includes day-ahead electricity price, real-time electricity price, medium and long-term transaction electricity price, power grid supply and demand data, and information on changes in electricity price policies and adjustments to transaction rules. The new energy construction auxiliary support raw data acquisition submodule is used to collect the new energy construction auxiliary support raw data, which includes the user site's usable area, roof and ground load-bearing capacity, regional meteorological data on sunlight, wind speed, and the enterprise's peak and off-peak season production plan data, and the inherent operating parameters of photovoltaic / wind power / energy storage equipment.

[0008] Optionally, the AI ​​electricity price prediction model uses a CNN convolutional neural network and a reflective large language model fusion algorithm to input the preprocessed standard data to obtain the time-of-use electricity price prediction data and electricity price fluctuation warning data; the AI ​​load prediction model uses an LSTM long short-term memory network algorithm to input the preprocessed standard data to obtain the electricity load prediction data and load fluctuation warning data; the electricity price and load fluctuation warning data include the electricity price fluctuation warning data and the load fluctuation warning data.

[0009] Optionally, the power consumption diagnostic module includes: The energy consumption characteristic diagnosis submodule is used to analyze the user's time-of-use electricity consumption, load factor, and peak-valley load difference based on the original electricity consumption data of the industrial and commercial users and the electricity load forecast data, and benchmark against the industry energy consumption benchmark value to obtain the quantitative diagnosis results and the existing electricity consumption pain point data. The quantitative diagnosis results are the user's energy consumption indicators and load utilization rate values ​​for each time period. The existing electricity consumption pain point data are problems such as low load utilization rate, abnormal electricity consumption, and poor load matching. The electricity cost diagnosis submodule is used to analyze the composition and proportion of users' time-of-use electricity purchase costs based on the raw dynamic data of the electricity market, the time-of-use electricity price prediction data, and the electricity price and load fluctuation early warning data, to obtain the quantitative diagnosis results and the existing electricity consumption pain point data. The quantitative diagnosis results are the overall electricity cost value and the value of the cost optimization space that can be explored. The existing electricity consumption pain point data are the problems of high load electricity consumption during peak electricity price periods, unreasonable market-based electricity purchase strategies, and high electricity costs. The power quality diagnosis submodule is used to monitor the core indicators of three-phase imbalance, harmonic content, and power factor in real time based on the original power consumption data of the industrial and commercial users, and obtain the quantitative diagnosis results and the existing power consumption pain point data. The quantitative diagnosis results are the measured values ​​of each power quality indicator and the quantitative amount of the annual power regulation penalty. The existing power consumption pain point data are power quality defects such as substandard power factor, three-phase imbalance, and excessive harmonics, as well as additional penalty losses. The new energy adaptability diagnosis submodule is used to obtain the quantitative diagnosis results and the existing electricity consumption pain point data based on the original electricity consumption data of industrial and commercial users, the time-of-use electricity price prediction data, the electricity load prediction data, and the electricity price and load fluctuation early warning data. The quantitative diagnosis results are the maximum installable potential values ​​of photovoltaic, wind power, and energy storage on the site, and the quantitative indicators of wind-solar-load matching. The existing electricity consumption pain point data are the shortcomings of users' underutilization of new energy resources, lack of wind-solar-storage coordinated configuration, and insufficient matching degree between load and new energy output.

[0010] Optionally, the optimal configuration module presets four layers of constraints and sets them in parallel. The constraints include: technical constraints, economic constraints, supply and demand balance constraints, and market constraints.

[0011] Optionally, the technical constraints are as follows: the installed capacity of new energy sources does not exceed the upper limit of site resources; the energy storage system has a charge / discharge depth of ≤80%, a round-trip efficiency of ≥85%, and a state of charge (SOC) that is constantly maintained at 20%-80%; and the overall microgrid grid-connected operation complies with the State Grid grid connection access standards.

[0012] Optionally, the economic constraint is that the initial investment and annual operation and maintenance costs of the system are controllable, the static investment payback period of the microgrid is ≤8 years, and the internal rate of return of the total investment is not lower than the benchmark rate of return of the new energy industry.

[0013] Optionally, the supply and demand balance constraint is to match the combined output of photovoltaic, wind, and energy storage with the real-time power load of users, with a wind and solar curtailment rate of ≤5% across the entire region, ensuring the reliability of power supply within users' premises.

[0014] Optionally, the market constraint is that the surplus electricity of the microgrid can support market-based surplus electricity grid connection and grid demand response services, and is compatible with the trading rules of the entire power market.

[0015] Optionally, the result output module outputs the time-of-use electricity price prediction curve, the electricity load prediction curve, the optimal installed capacity, the layout scheme and the photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations, a full-dimensional electricity consumption diagnosis and analysis report, full life cycle cost and benefit calculation data, carbon emission reduction benefits and carbon asset value accounting data, and a project implementation risk prevention and control plan.

[0016] The present invention discloses the following technical effects: This invention relies on a multi-source data acquisition, cleaning, verification, and normalization preprocessing mechanism, utilizing raw data on basic electricity consumption of industrial and commercial users, dynamic raw data on the electricity market, and raw data supporting new energy construction. Combined with AI-powered electricity price prediction and load prediction models, it achieves high-precision dynamic prediction of electricity market prices and user electricity load across multiple time scales. This effectively solves the shortcomings of traditional technologies, such as static analysis, low prediction accuracy, and inability to adapt to dynamic fluctuations in electricity market prices. By integrating raw data on basic electricity consumption of industrial and commercial users with time-of-use electricity price prediction data, electricity load prediction data, and price and load fluctuation early warning data, this invention completes dynamic electricity consumption diagnosis across four dimensions: energy consumption characteristics, electricity cost, power quality, and new energy adaptability. It accurately outputs quantitative diagnostic results and data on existing electricity consumption pain points, providing precise data support for microgrid optimization. This invention integrates preprocessed standard data, AI prediction data, quantitative diagnostic results, and existing electricity consumption pain point data to solve for the optimal installed capacity and layout scheme of integrated photovoltaic, wind power, and energy storage. It also generates a photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations, overcoming the shortcomings of traditional microgrid configuration schemes, such as weak targeting, poor wind-solar-storage synergy, and inability to optimize solutions to users' actual electricity consumption pain points. In the context of dynamic electricity market pricing, this invention achieves intelligent and precise microgrid configuration with optimal electricity costs as its core, effectively reducing the total lifecycle electricity costs and equipment investment costs for industrial and commercial users, significantly improving the self-consumption rate of renewable energy and the reliability of power supply on the user side, and simultaneously enhancing carbon emission reduction benefits. It is highly adaptable to the actual application needs of industrial and commercial users for electricity cost reduction and low-carbon transformation of renewable energy. Attached Figure Description

[0017] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the overall modules of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, this invention provides a microgrid configuration system for optimal electricity cost in an electricity market based on AI prediction. The system includes: The data acquisition module is used to collect raw data and perform preprocessing such preprocessing, including cleaning, verification, and normalization, as well as removing invalid and abnormal data to obtain preprocessed standard data. The raw data includes basic raw data on electricity consumption of industrial and commercial users, dynamic raw data on the electricity market, and auxiliary support raw data for new energy construction. The AI ​​prediction module is used to build AI electricity price prediction models and AI load prediction models. Based on the preprocessed standard data, it completes dynamic predictions of electricity market prices and user electricity loads at multiple time scales in the future, and obtains time-of-use electricity price prediction data, electricity load prediction data, and electricity price and load fluctuation early warning data. The electricity consumption diagnosis module is used to perform dynamic electricity consumption diagnosis based on the original electricity consumption data of industrial and commercial users, the time-of-use electricity price prediction data, the electricity load prediction data, and the electricity price and load fluctuation early warning data. It can also output quantitative diagnosis results and existing electricity consumption pain point data. The optimal configuration module is used to output the optimal installed capacity and layout scheme of photovoltaic, wind power and energy storage integration based on the preprocessed standard data, the time-of-use electricity price prediction data, the electricity load prediction data, the electricity price and load fluctuation early warning data, the quantitative diagnosis results and the existing electricity consumption pain point data, and at the same time generate a photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations. The results output module is used to visualize the time-of-use electricity price forecast data, the electricity load forecast data, the electricity price and load fluctuation early warning data, the quantitative diagnostic results, the existing electricity consumption pain point data, the optimal installed capacity, the layout scheme, and the photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations. It also generates an exportable standardized configuration diagnostic report and supports manual parameter adjustment and scheme recalculation.

[0021] First, the data acquisition module collects three types of raw data in batches: basic electricity consumption data of industrial and commercial users, dynamic raw data of the electricity market, and raw data supporting new energy construction. All collected raw data undergoes unified cleaning, verification, and normalization preprocessing to remove invalid and abnormal data caused by power outages, equipment failures, and signal interference, resulting in standardized and structured preprocessed data. Then, the AI ​​prediction module, based on the preprocessed standard data, uses its built-in AI electricity price prediction model and AI load prediction model to perform dynamic predictions of electricity market prices and user electricity loads at multiple time scales, accurately outputting time-of-use price prediction data, electricity load prediction data, and price and load fluctuation warning data. The electricity consumption diagnosis module retrieves the basic electricity consumption data of industrial and commercial users and the three types of prediction data output by the AI ​​prediction module, conducting dynamic electricity consumption diagnosis from four dimensions: energy consumption characteristics, electricity cost, power quality, and new energy adaptability, quantitatively outputting the diagnostic results and existing electricity consumption pain point data. The optimal configuration module integrates preprocessed standard data, AI-predicted full-data, diagnostic quantitative results, and electricity consumption pain point data to intelligently solve and output the optimal installed capacity and layout scheme for integrated photovoltaic, wind power, and energy storage. It also adapts to the dynamic electricity market pricing and load fluctuation characteristics, generating corresponding photovoltaic-storage-wind collaborative operation strategies. Finally, the results output module visualizes the core data and optimization schemes throughout the entire process, covering time-of-use price forecast data, electricity load forecast data, price and load fluctuation early warning data, quantitative diagnostic results, existing electricity consumption pain point data, optimal installed capacity, layout scheme, and photovoltaic-storage-wind collaborative operation strategies. It automatically generates an exportable standardized configuration diagnostic report and supports manual parameter adjustment and secondary recalculation of the scheme, meeting the personalized configuration optimization needs of different industrial and commercial users.

[0022] Furthermore, the data acquisition module includes: The industrial and commercial user electricity consumption basic raw data acquisition submodule is used to collect the industrial and commercial user electricity consumption basic raw data, which includes active power, reactive power, voltage, current, time-of-use electricity consumption, power factor, and load factor. The electricity market dynamic raw data acquisition submodule is used to collect the electricity market dynamic raw data, which includes day-ahead electricity price, real-time electricity price, medium and long-term transaction electricity price, power grid supply and demand data, and information on changes in electricity price policies and adjustments to transaction rules. The new energy construction auxiliary support raw data acquisition submodule is used to collect the new energy construction auxiliary support raw data, which includes the user site's usable area, roof and ground load-bearing capacity, regional meteorological data on sunlight, wind speed, and the enterprise's peak and off-peak season production plan data, and the inherent operating parameters of photovoltaic / wind power / energy storage equipment.

[0023] The data acquisition module is specifically divided into three functional sub-modules, each responsible for the accurate acquisition of multi-source raw data. The industrial and commercial user electricity basic raw data acquisition sub-module uses high-precision power acquisition terminals deployed in all load branch circuits of industrial and commercial user distribution lines, production workshops, and office areas to collect core electricity parameters in real time, such as active power, reactive power, voltage, current, time-of-use electricity consumption, power factor, and load factor. The data acquisition frequency is set to once every 15 minutes to ensure the real-time, complete, and accurate nature of the electricity data. The electricity market dynamic raw data acquisition sub-module connects to the provincial electricity market trading platform in real time through an encrypted data interface, collecting day-ahead electricity prices, real-time electricity prices, medium- and long-term trading prices, grid supply and demand data, electricity price policy changes, and trading rule adjustments around the clock, synchronizing with real-time changes in the electricity market. The raw data acquisition submodule for new energy construction support collects data on the usable area of ​​user sites, roof and ground load-bearing parameters, regional solar radiation, wind speed, and meteorological data, enterprise production schedules for peak and off-peak seasons, and inherent operating parameters of photovoltaic, wind power, and energy storage equipment through site survey equipment, meteorological monitoring equipment, and enterprise production ledgers. This provides fundamental support for subsequent optimization of new energy configurations. The raw data collected by each submodule is uniformly aggregated at the backend of the data acquisition module for centralized preprocessing.

[0024] Furthermore, the AI ​​electricity price prediction model uses a CNN convolutional neural network and a reflective large language model fusion algorithm to input the preprocessed standard data to obtain the time-of-use electricity price prediction data and electricity price fluctuation warning data; the AI ​​load prediction model uses an LSTM long short-term memory network algorithm to input the preprocessed standard data to obtain the electricity load prediction data and load fluctuation warning data; the electricity price and load fluctuation warning data include the electricity price fluctuation warning data and the load fluctuation warning data.

[0025] The AI ​​prediction module comprises two dedicated prediction models, employing differentiated algorithms to achieve accurate time-series predictions. Data input and output are strictly matched to the overall system data link. Specifically, the AI ​​electricity price prediction model uses a fusion algorithm of CNN convolutional neural networks and a reflective large language model. Using preprocessed standard data as the sole core input, it mines the time-series changes in electricity market prices and the characteristics of supply and demand relationships, accurately outputting time-of-use electricity price predictions for multiple time scales. Simultaneously, it outputs electricity price fluctuation warning data based on the magnitude of price fluctuations and policy changes. The AI ​​load prediction model uses the LSTM long short-term memory network algorithm. Leveraging the advantages of LSTM's long short-term memory, it learns from preprocessed standard data the historical electricity consumption patterns of users, seasonal weather influences, and production scheduling fluctuation characteristics, outputting future electricity load prediction data for multiple time scales. It also outputs load fluctuation warning data based on sudden load changes. Finally, the electricity price fluctuation warning data and load fluctuation warning data are integrated to form complete electricity price and load fluctuation warning data, providing accurate pre-prediction support for electricity consumption diagnosis and optimized allocation.

[0026] Optionally, the power consumption diagnostic module includes: The energy consumption characteristic diagnosis submodule is used to analyze the user's time-of-use electricity consumption, load factor, and peak-valley load difference based on the original electricity consumption data of the industrial and commercial users and the electricity load forecast data, and benchmark against the industry energy consumption benchmark value to obtain the quantitative diagnosis results and the existing electricity consumption pain point data. The quantitative diagnosis results are the user's energy consumption indicators and load utilization rate values ​​for each time period. The existing electricity consumption pain point data are problems such as low load utilization rate, abnormal electricity consumption, and poor load matching. The electricity cost diagnosis submodule is used to analyze the composition and proportion of users' time-of-use electricity purchase costs based on the raw dynamic data of the electricity market, the time-of-use electricity price prediction data, and the electricity price and load fluctuation early warning data, to obtain the quantitative diagnosis results and the existing electricity consumption pain point data. The quantitative diagnosis results are the overall electricity cost value and the value of the cost optimization space that can be explored. The existing electricity consumption pain point data are the problems of high load electricity consumption during peak electricity price periods, unreasonable market-based electricity purchase strategies, and high electricity costs. The power quality diagnosis submodule is used to monitor the core indicators of three-phase imbalance, harmonic content, and power factor in real time based on the original power consumption data of the industrial and commercial users, and obtain the quantitative diagnosis results and the existing power consumption pain point data. The quantitative diagnosis results are the measured values ​​of each power quality indicator and the quantitative amount of the annual power regulation penalty. The existing power consumption pain point data are power quality defects such as substandard power factor, three-phase imbalance, and excessive harmonics, as well as additional penalty losses. The new energy adaptability diagnosis submodule is used to obtain the quantitative diagnosis results and the existing electricity consumption pain point data based on the original electricity consumption data of industrial and commercial users, the time-of-use electricity price prediction data, the electricity load prediction data, and the electricity price and load fluctuation early warning data. The quantitative diagnosis results are the maximum installable potential values ​​of photovoltaic, wind power, and energy storage on the site, and the quantitative indicators of wind-solar-load matching. The existing electricity consumption pain point data are the shortcomings of users' underutilization of new energy resources, lack of wind-solar-storage coordinated configuration, and insufficient matching degree between load and new energy output.

[0027] The electricity consumption diagnosis module comprises four sub-modules, which strictly rely on limited input data to complete four-dimensional diagnosis, outputting corresponding quantitative diagnostic results and existing electricity consumption pain point data. The energy consumption characteristic diagnosis sub-module, based on the original electricity consumption data and predicted load data of industrial and commercial users, compares the user's actual load curve with the predicted load curve, analyzes time-of-use electricity consumption, load factor, and peak-valley load difference, benchmarks against industry energy consumption benchmarks, and quantifies energy consumption indicators and load utilization rates for each time period. It also identifies electricity consumption pain points such as low load utilization, abnormal energy consumption, and poor load matching. The electricity cost diagnosis sub-module combines dynamic original data of the electricity market, time-of-use electricity price prediction data, and electricity price and load fluctuation early warning data to break down the composition and proportion of users' time-of-use electricity purchase costs, quantifies the overall electricity consumption cost and the potential for cost optimization, and identifies pain points such as high load electricity consumption at peak prices, unreasonable electricity purchase strategies, and high electricity costs. The power quality diagnosis submodule, based on raw data of industrial and commercial users' electricity consumption, monitors three-phase imbalance, harmonic content, and power factor in real time. It quantifies the measured values ​​of each power quality indicator and the annual power regulation penalty amount, identifying power quality pain points such as substandard power factor, three-phase imbalance, excessive harmonics, and additional penalty losses. The renewable energy adaptability diagnosis submodule, relying on raw data of industrial and commercial users' electricity consumption, time-of-use price forecast data, electricity load forecast data, and price and load fluctuation early warning data, quantifies and calculates the maximum installed capacity potential of wind, solar, and energy storage at the site, as well as the wind-solar-load matching degree index. It identifies adaptability shortcomings such as underutilization of renewable energy resources, lack of coordinated wind, solar, and energy storage configuration, and insufficient matching degree between load and renewable energy output.

[0028] Furthermore, the optimal configuration module presets four layers of constraints and sets them in parallel. These constraints include: technical constraints, economic constraints, supply and demand balance constraints, and market constraints.

[0029] In the optimization process, the optimal configuration module pre-sets four layers of parallel constraints. All constraints participate in the iterative calculation simultaneously, jointly limiting the rationality, feasibility, and economy of the microgrid configuration scheme. The four layers of constraints specifically include technical constraints, economic constraints, supply and demand balance constraints, and market constraints. These four types of constraints cooperate with each other without conflict, and comprehensively constrain the installed capacity, layout, and operation strategies of photovoltaic, wind power, and energy storage from four dimensions: equipment technical parameters, project investment returns, system supply and demand balance, and electricity market adaptability. This avoids problems such as technically infeasible configuration schemes, poor economic returns, unstable power supply, and inability to adapt to market trading rules, ensuring that the final output microgrid configuration scheme meets the actual needs of industrial and commercial users.

[0030] Furthermore, the technical constraints are as follows: the installed capacity of new energy sources does not exceed the upper limit of site resources; the energy storage system has a charge / discharge depth of ≤80%, a round-trip efficiency of ≥85%, and a state of charge (SOC) that is constantly maintained between 20% and 80%; and the overall microgrid grid-connected operation complies with the State Grid grid connection access standards.

[0031] Specifically, the installed capacity of new energy sources is strictly limited to not exceeding the resource ceiling corresponding to the usable area and load-bearing capacity of the user's site, eliminating the problem of resource allocation exceeding the site's capacity. For energy storage systems, the depth of charge and discharge is strictly controlled to be ≤80%, the energy storage round-trip efficiency is ≥85%, and the state of charge (SOC) is constantly maintained within the range of 20%-80%, effectively extending the service life of energy storage equipment and ensuring the stable operation of the energy storage system. At the same time, the grid connection parameters, wiring methods, and protection strategies of all photovoltaic, wind power, and energy storage equipment strictly comply with the State Grid's microgrid grid connection access specifications, ensuring that the microgrid system can be safely and compliantly connected to the grid without any grid connection technical risks.

[0032] Furthermore, the economic constraints are that the initial investment and annual operation and maintenance costs of the system are controllable, the static investment payback period of the microgrid is ≤8 years, and the internal rate of return of the total investment is not lower than the benchmark rate of return of the new energy industry.

[0033] The system combines equipment procurement costs, engineering construction costs, and annual operation and maintenance costs to calculate the investment and returns throughout the project's entire lifecycle, strictly controlling the overall investment and operation and maintenance costs to remain within the user's control. During the optimization process, the static investment payback period of the microgrid is forcibly limited to ≤8 years, while ensuring that the internal rate of return on total investment is not lower than the benchmark rate of return in the new energy industry. This effectively avoids the problems of excessively long payback periods and unmet returns, ensuring that industrial and commercial users investing in microgrid projects achieve good economic returns.

[0034] Furthermore, the supply and demand balance constraint is to match the combined output of photovoltaic, energy storage and wind power with the real-time power load of users, with the wind curtailment rate of photovoltaic and solar power in the entire region being ≤5%, ensuring the reliability of power supply within users' premises.

[0035] The system dynamically matches the combined output of photovoltaic and wind power with the real-time electricity load of users. Through the charging and discharging regulation of the energy storage system, it smooths out fluctuations in renewable energy output and balances the system's supply and demand in real time. At the same time, it strictly controls the curtailment rate of wind and solar power across the entire region to ≤5%, maximizes the local consumption rate of renewable energy, avoids waste of renewable energy resources, continuously ensures the reliability of power supply to users' internal electricity loads, and achieves dynamic adaptation between renewable energy output and user electricity loads.

[0036] Furthermore, the market constraint is that the surplus electricity of the microgrid can support market-based surplus electricity grid connection and grid demand response services, and is compatible with the trading rules of the entire power market.

[0037] The system's microgrid configuration supports market-based trading of surplus electricity, and can participate in grid demand response services based on electricity market price fluctuations and grid dispatch needs. The microgrid's operation strategy, electricity trading rules, and price adaptation logic all align with the latest trading rules and policy requirements of the global electricity market, fully leveraging the market-based trading mechanism to further reduce user electricity costs and increase project revenue.

[0038] Furthermore, the result output module outputs time-of-use electricity price prediction curves, electricity load prediction curves, the optimal installed capacity, the layout scheme and the photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations, a full-dimensional electricity consumption diagnosis and analysis report, full life cycle cost and benefit calculation data, carbon emission reduction benefits and carbon asset value accounting data, and a project implementation risk prevention and control plan.

[0039] The results output module is responsible for the visualization and data output of the core outcomes throughout the system's entire process. Specific outputs include: time-of-use electricity price prediction curves and electricity load prediction curves generated based on AI-predicted data; optimal installed capacity and layout schemes obtained from the optimal configuration module; and a photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations. Simultaneously, the module automatically integrates and generates comprehensive electricity consumption diagnostic analysis reports, microgrid lifecycle cost-benefit calculation data, project carbon emission reduction benefits, and carbon asset value accounting data. It also provides a risk control plan for project implementation. All outputs can be visualized and exported in batches, providing complete and intuitive data support for user project decision-making, implementation, and benefit evaluation.

[0040] Example This embodiment first completes the collection and standardized preprocessing of multi-source data across the entire domain through a data acquisition module, strictly corresponding to the three types of raw data defined in the claims: basic raw data of industrial and commercial user electricity consumption, dynamic raw data of the electricity market, and raw data supporting new energy construction. This embodiment deploys 0.5S-level high-precision power acquisition terminals at the enterprise's high-voltage incoming line, low-voltage main switchgear, and load branch circuits in various production workshops and offices, with a collection frequency set to 15 minutes / time, continuously collecting basic raw data of industrial and commercial user electricity consumption such as active power, reactive power, voltage, current, time-of-use electricity consumption, power factor, and load factor. Through an encrypted official interface connected to the provincial power trading platform, it collects dynamic raw data of the electricity market in real time, including day-ahead electricity prices, real-time electricity prices, medium- and long-term trading prices, grid supply and demand data, and information on changes in electricity price policies and adjustments to trading rules. Through a third-party meteorological platform and enterprise site surveys and production ledgers, it collects regional solar radiation and wind speed meteorological data, enterprise production peak and off-peak season plans, usable site area, roof and ground load-bearing capacity, and inherent operating parameters of wind, solar, and energy storage equipment, among other raw data supporting new energy construction. The data acquisition module performs cleaning, verification, and normalization preprocessing on all raw data to remove invalid and abnormal data caused by power outages, maintenance, signal anomalies, and equipment failures. In this embodiment, the effective data coverage rate reaches 99.2%, and finally generates preprocessed standard data with standardized format and uniform dimensions that can be directly used for model calculations.

[0041] All preprocessed standard data is input into the AI ​​prediction module. This module strictly adheres to the specifications of the claims and is configured with dual AI prediction models: an AI electricity price prediction model fused with a CNN convolutional neural network and a reflective large language model, and an LSTM long short-term memory network AI load prediction model. The AI ​​electricity price prediction model uses the preprocessed standard data as its core input, mining the temporal correlation features of electricity market supply and demand, meteorology, and policies to complete dynamic predictions of short-term electricity prices for the next 72 hours and long-term electricity prices for the next 30 days. In this embodiment, the electricity price prediction accuracy reaches 86.5%, simultaneously identifying the magnitude of electricity price fluctuations and outputting electricity price fluctuation warning data. The AI ​​load prediction model leverages the advantages of LSTM temporal memory to fit enterprise production schedules, seasonal weather, and peak / off-peak load variation patterns, completing multi-timescale electricity load predictions. In this embodiment, the load prediction accuracy reaches 91.3%, accurately capturing peak weekday loads and stable nighttime loads, and outputting load fluctuation warning data. Finally, the two types of warning data are integrated to form complete electricity price and load fluctuation warning data, providing high-precision and forward-looking data support for backend diagnosis and optimization.

[0042] After completing AI predictions, the system conducts dynamic electricity consumption diagnosis across four dimensions through its electricity consumption diagnosis module. The diagnostic inputs strictly adhere to the limitations of the claims, including basic raw data on industrial and commercial user electricity consumption, time-of-use electricity price prediction data, electricity load prediction data, and electricity price and load fluctuation warning data. The final output is a unified quantitative diagnostic result and data on existing electricity consumption pain points. In terms of energy consumption characteristic diagnosis, the system compares the company's actual raw load curve with the AI-predicted load curve, quantifying that the company's average transformer load rate is 68% and its overall load rate is 72%, lower than the industry benchmark of 75%, thus forming a quantitative diagnostic result for energy consumption indicators. Simultaneously, it identifies electricity consumption pain points such as low load utilization and insufficient load matching. In terms of electricity cost diagnosis, the system combines raw electricity consumption data, dynamic electricity price prediction data, and fluctuation warning data to quantitatively calculate the company's annual electricity cost of 4.8 million yuan, peak-hour electricity cost accounting for 62%, and an annual cost optimization space of 1.2 million yuan. This identifies the core pain points of high-load production during peak electricity price periods, unreasonable market-based electricity purchase strategies, and high electricity costs. In terms of power quality diagnosis, the system, based on real-time raw electricity consumption data monitoring, determined that the company's average power factor was 0.88, failing to meet the State Grid's access standards, resulting in an annual power adjustment fine of approximately 80,000 yuan. This clearly identified the pain points of additional economic losses caused by substandard power factor and power quality defects. Regarding renewable energy adaptability diagnosis, the system combined raw electricity consumption patterns with predicted fluctuation characteristics to quantitatively calculate the company's maximum potential installed capacity for rooftop and ground-mounted wind, solar, and energy storage, as well as the matching degree of wind and solar loads. The diagnosis revealed that the company possessed excellent conditions for renewable energy transformation, but suffered from shortcomings such as unutilized renewable energy resources, lack of coordinated wind, solar, and energy storage configuration, and insufficient matching degree between load and renewable energy output.

[0043] After diagnosis, the optimal configuration module integrates preprocessed standard data, time-of-use electricity price forecast data, electricity load forecast data, electricity price and load fluctuation early warning data, quantitative diagnostic results, and existing electricity consumption pain point data. It then uses a multi-objective genetic algorithm to iteratively optimize based on four layers of parallel constraints. This embodiment strictly adheres to the constraints defined in the claims: technical constraints limit wind and solar installed capacity to the upper limit of site resources, energy storage charge / discharge depth ≤80%, round-trip efficiency ≥85%, SOC maintained at 20%-80%, and grid connection conforming to State Grid standards; economic constraints strictly control the project investment payback period to ≤8 years and the internal rate of return to be higher than the industry benchmark; supply and demand balance constraints ensure that the combined output of solar, energy storage, and wind power matches real-time load, and the overall wind and solar curtailment rate is ≤5%; market constraints ensure that surplus electricity can be market-based and adapted to demand response and electricity market trading rules. This embodiment prioritizes minimizing the total cost of electricity over its entire lifecycle (weight 0.6), with carbon reduction and power supply reliability as secondary objectives (each 0.2). It sets the population size to 80, the number of iterations to 80, the crossover probability to 0.8, and the mutation probability to 0.03. The iterative solution outputs the optimal configuration: 2.2MWp photovoltaic capacity, 0.4MW wind power capacity, and 0.8MW / 1.6MWh energy storage. Simultaneously, it generates a photovoltaic-storage-wind synergistic operation strategy adapted to dynamic electricity prices and load fluctuations: during the day, wind and solar power are prioritized for local consumption, with surplus electricity used for energy storage and charging; during peak electricity price periods, energy storage discharges to reduce costs and peak shaving, while during off-peak periods, it utilizes lower-priced grid electricity for supplemental power; at night, it relies on wind power for supplemental power and energy storage to smooth out fluctuations, achieving optimal dynamic costs throughout the entire time period.

[0044] The final output module provides a fully visualized display of the entire process and delivers the results. The system visualizes time-of-use electricity price forecast curves, electricity load forecast curves, price and load fluctuation warnings, four-dimensional quantitative diagnostic results, existing electricity consumption pain points for the enterprise, optimal installed capacity, site layout plans, and dynamic collaborative operation strategies. It automatically generates exportable standardized configuration diagnostic reports, including comprehensive diagnostic reports, full lifecycle cost-benefit calculations, carbon emission reduction benefit calculations, and implementation risk control plans. After the implementation of this solution, the enterprise's annual electricity cost decreased to 3.52 million yuan, resulting in annual electricity cost savings of 1.28 million yuan. The static investment payback period is 6.8 years, the internal rate of return on total investment is 12.3%, annual carbon emission reduction is 1600 tons of carbon dioxide, the power factor is improved to 0.92, power regulation penalties are completely eliminated, and power supply reliability is improved to 99.5%. Simultaneously, the system supports manual parameter adjustments and secondary recalculation of the solution, allowing for continuous iteration and optimization of configuration and operation strategies based on changes in electricity market policies and enterprise production adjustments, maintaining optimal electricity costs in the long term.

[0045] In summary, this embodiment fully replicates the closed-loop technical architecture of the present invention, thoroughly solving the shortcomings of traditional technologies such as inability to adapt to market-based dynamic electricity prices, low prediction accuracy, disconnect between diagnosis and configuration, poor synergy between wind, solar and energy storage, and single optimization objective. Through multi-source data standardization processing, high-precision prediction using dual AI models, dynamic diagnosis coupled with truth value and prediction, and intelligent optimization with multiple constraints and objectives, it achieves precise, intelligent, and cost-optimized microgrid configuration for industrial and commercial users, while taking into account cost reduction and efficiency improvement as well as low-carbon transformation. It has strong engineering feasibility, wide applicability, and significant economic and social benefits.

[0046] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A microgrid configuration system for optimal electricity cost in an electricity market based on AI prediction, characterized in that, The system includes: The data acquisition module is used to collect raw data and perform preprocessing such preprocessing, including cleaning, verification, and normalization, as well as removing invalid and abnormal data to obtain preprocessed standard data. The raw data includes basic raw data on electricity consumption of industrial and commercial users, dynamic raw data on the electricity market, and auxiliary support raw data for new energy construction. The AI ​​prediction module is used to build AI electricity price prediction models and AI load prediction models. Based on the preprocessed standard data, it completes dynamic predictions of electricity market prices and user electricity loads at multiple time scales in the future, and obtains time-of-use electricity price prediction data, electricity load prediction data, and electricity price and load fluctuation early warning data. The electricity consumption diagnosis module is used to perform dynamic electricity consumption diagnosis based on the original electricity consumption data of industrial and commercial users, the time-of-use electricity price prediction data, the electricity load prediction data, and the electricity price and load fluctuation early warning data. It can also output quantitative diagnosis results and existing electricity consumption pain point data. The optimal configuration module is used to output the optimal installed capacity and layout scheme of photovoltaic, wind power and energy storage integration based on the preprocessed standard data, the time-of-use electricity price prediction data, the electricity load prediction data, the electricity price and load fluctuation early warning data, the quantitative diagnosis results and the existing electricity consumption pain point data, and at the same time generate a photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations. The results output module is used to visualize the time-of-use electricity price forecast data, the electricity load forecast data, the electricity price and load fluctuation early warning data, the quantitative diagnostic results, the existing electricity consumption pain point data, the optimal installed capacity, the layout scheme, and the photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations. It also generates an exportable standardized configuration diagnostic report and supports manual parameter adjustment and scheme recalculation.

2. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market as described in claim 1, characterized in that, The data acquisition module includes: The industrial and commercial user electricity consumption basic raw data acquisition submodule is used to collect the industrial and commercial user electricity consumption basic raw data, which includes active power, reactive power, voltage, current, time-of-use electricity consumption, power factor, and load factor. The electricity market dynamic raw data acquisition submodule is used to collect the electricity market dynamic raw data, which includes day-ahead electricity price, real-time electricity price, medium and long-term transaction electricity price, power grid supply and demand data, and information on changes in electricity price policies and adjustments to transaction rules. The new energy construction auxiliary support raw data acquisition submodule is used to collect the new energy construction auxiliary support raw data, which includes the user site's usable area, roof and ground load-bearing capacity, regional meteorological data on sunlight, wind speed, and the enterprise's peak and off-peak season production plan data, and inherent operating parameters of photovoltaic / wind power / energy storage equipment.

3. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market as described in claim 1, characterized in that, The AI ​​electricity price prediction model uses a CNN convolutional neural network and a reflective large language model fusion algorithm to input the preprocessed standard data to obtain the time-of-use electricity price prediction data and the electricity price fluctuation early warning data. The AI ​​load forecasting model uses the LSTM long short-term memory network algorithm to input the preprocessed standard data to obtain the electricity load forecasting data and load fluctuation early warning data. The electricity price and load fluctuation early warning data includes the electricity price fluctuation early warning data and the load fluctuation early warning data.

4. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market as described in claim 1, characterized in that, The power consumption diagnostic module includes: The energy consumption characteristic diagnosis submodule is used to analyze the user's time-of-use electricity consumption, load factor, and peak-valley load difference based on the original electricity consumption data of the industrial and commercial users and the electricity load forecast data, and benchmark against the industry energy consumption benchmark value to obtain the quantitative diagnosis results and the existing electricity consumption pain point data. The quantitative diagnosis results are the user's energy consumption indicators and load utilization rate values ​​for each time period. The existing electricity consumption pain point data are problems such as low load utilization rate, abnormal electricity consumption, and poor load matching. The electricity cost diagnosis submodule is used to analyze the composition and proportion of users' time-of-use electricity purchase costs based on the raw dynamic data of the electricity market, the time-of-use electricity price prediction data, and the electricity price and load fluctuation early warning data, to obtain the quantitative diagnosis results and the existing electricity consumption pain point data. The quantitative diagnosis results are the overall electricity cost value and the value of the cost optimization space that can be explored. The existing electricity consumption pain point data are the problems of high load electricity consumption during peak electricity price periods, unreasonable market-based electricity purchase strategies, and high electricity costs. The power quality diagnosis submodule is used to monitor the core indicators of three-phase imbalance, harmonic content, and power factor in real time based on the original power consumption data of the industrial and commercial users, and obtain the quantitative diagnosis results and the existing power consumption pain point data. The quantitative diagnosis results are the measured values ​​of each power quality indicator and the quantitative amount of the annual power regulation penalty. The existing power consumption pain point data are power quality defects such as substandard power factor, three-phase imbalance, and excessive harmonics, as well as additional penalty losses. The new energy adaptability diagnosis submodule is used to obtain the quantitative diagnosis results and the existing electricity consumption pain point data based on the original electricity consumption data of industrial and commercial users, the time-of-use electricity price prediction data, the electricity load prediction data, and the electricity price and load fluctuation early warning data. The quantitative diagnosis results are the maximum installable potential values ​​of photovoltaic, wind power, and energy storage on the site, and the quantitative indicators of wind-solar-load matching. The existing electricity consumption pain point data are the shortcomings of users' underutilization of new energy resources, lack of wind-solar-storage coordinated configuration, and insufficient matching degree between load and new energy output.

5. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market as described in claim 1, characterized in that, The optimal configuration module presets four layers of constraints and sets them in parallel. The constraints include: technical constraints, economic constraints, supply and demand balance constraints, and market constraints.

6. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market according to claim 5, characterized in that, The technical constraint is that the installed capacity of new energy sources shall not exceed the upper limit of site resources; The energy storage system has a charge / discharge depth of ≤80%, a round-trip efficiency of ≥85%, and a state of charge (SOC) that is constantly maintained between 20% and 80%. The overall microgrid operation complies with the State Grid's grid connection access standards.

7. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market according to claim 5, characterized in that, The economic constraints are that the initial investment and annual operation and maintenance costs of the system are controllable, the static investment payback period of the microgrid is ≤8 years, and the internal rate of return of the total investment is not lower than the benchmark rate of return of the new energy industry.

8. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market according to claim 5, characterized in that, The supply and demand balance constraint is to match the combined output of photovoltaic, energy storage and wind power with the real-time power load of users, with the wind curtailment rate of the entire area ≤5%, ensuring the reliability of the power supply inside the user's premises.

9. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market according to claim 5, characterized in that, The market constraints are that the surplus electricity of the microgrid can support market-based surplus electricity grid connection and grid demand response services, and be compatible with the trading rules of the entire power market.

10. The microgrid configuration system for optimal electricity cost in an AI-predicted electricity market according to claim 1, characterized in that, The output module outputs the time-of-use electricity price prediction curve, the electricity load prediction curve, the optimal installed capacity, the layout scheme and the photovoltaic-storage-wind collaborative operation strategy adapted to dynamic electricity prices and load fluctuations, the full-dimensional electricity consumption diagnosis and analysis report, the full life cycle cost and benefit calculation data, the carbon emission reduction benefits and carbon asset value accounting data, and the project implementation risk prevention and control plan.

Images