A multi-user load forecasting and electricity purchasing decision method and system for a green electricity direct connection project main responsibility unit

Abstract

The application discloses a multi-user load forecasting and electricity purchasing decision method and system for a green electricity direct connection project responsible unit, and belongs to the technical field of power system optimization dispatching and energy management. The method comprises the following steps: collecting the industry types, historical electricity consumption data, production plans and meteorological data of each user in the project responsible unit; classifying the users according to the industry types and respectively establishing load forecasting models to output the load forecasting values of each user in a future preset time period; acquiring the time-of-use electricity price of a large power grid, a green electricity premium, internal photovoltaic / wind power forecasting output, an energy storage system state and a unit electricity cost; constructing an optimization model with the minimum overall electricity purchasing cost of the project responsible unit as a target, and solving an electricity purchasing strategy, an energy storage dispatching strategy, a user electricity selling price and a green electricity quota; and pushing the optimization result to an edge computing device for execution, and rolling correction of subsequent plans every 15 minutes. The application also provides a system for implementing the method and a computer readable storage medium. The application fills the gap of a special electricity purchasing optimization tool for the project responsible unit, a new market subject, and can reduce the overall electricity purchasing cost by 10%-20%.

Description

Technical Field

[0001] This invention belongs to the field of power system optimization scheduling and energy management technology, specifically involving a multi-user load forecasting and power purchase decision-making method and system for the main responsible unit of green electricity direct connection projects. Background Technology

[0002] With the advancement of the national policy on direct green electricity connection for multiple users, a new type of market entity, the "project lead unit," has emerged. The project lead unit needs to simultaneously undertake multiple functions, including bundled electricity purchase, electricity sales, power generation, distribution, energy storage operation, and shared energy storage, similar to a power sales company or integrated energy service provider. Unlike traditional power sales companies, project lead units manage diverse user types with significantly different load characteristics, including industrial enterprises (high volatility), commercial office buildings (cyclical), and data centers (stable but with high demand). Furthermore, they need to comprehensively consider the output fluctuations of internal distributed photovoltaic and wind power, the charging and discharging scheduling of energy storage systems, and policy requirements for green electricity consumption.

[0003] However, existing technologies lack dedicated load forecasting and power purchase decision-making tools for this type of entity. Traditional load forecasting methods only target single users or grid nodes, failing to handle the heterogeneous load characteristics of multi-user systems. Their forecasting accuracy is insufficient to meet the needs of project owners for refined power purchase decisions. Power purchase strategies are often based on fixed peak-valley prices or empirical rules, failing to comprehensively optimize internal photovoltaic output, wind power forecasting, energy storage regulation, and multi-user green electricity allocation. This results in higher overall power purchase costs and lower green electricity consumption rates, failing to fully leverage the scale advantages and resource allocation capabilities of the project owner as a unified power purchase entity.

[0004] Therefore, there is an urgent need for a load forecasting and power purchase decision-making method and system that can be oriented towards the project's responsible unit, adapt to the heterogeneous load characteristics of multiple users, integrate electricity price signals and green electricity consumption requirements, and minimize the overall power purchase cost. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-user load forecasting and power purchase decision-making method and system for the main responsible unit of green electricity direct connection projects, so as to solve the problems in the prior art of lacking dedicated power purchase optimization tools for the main responsible unit of the project, low accuracy of multi-user load forecasting, and power purchase strategy that does not comprehensively optimize green electricity allocation and energy storage scheduling.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a multi-user load forecasting and power purchase decision-making method for the main responsible unit of a green electricity direct connection project, comprising the following steps:

[0008] Step S1: Collect industry type, historical electricity consumption data, production plans and meteorological data of each user within the project's responsible unit.

[0009] Step S2: Classify users based on the industry type, and establish a load forecasting model for each user category, outputting load forecast values ​​for a preset future period. Preferably, the load forecasting model uses a Long Short-Term Memory network or a temporal convolutional network, and combines it with quantile regression to simultaneously output the load forecast interval for assessing forecast uncertainty. Different model parameters are used for users in different industry types: industrial enterprise users focus on capturing the impact of production shifts and order fluctuations, commercial office building users focus on capturing work-rest patterns and seasonal effects, and data center users focus on capturing the correlation between IT equipment load and cooling system.

[0010] Step S3: Obtain the time-of-use electricity price of the main power grid, the green electricity premium, the predicted output of internal photovoltaic / wind power, the status of energy storage systems, and the cost per kilowatt-hour.

[0011] Step S4: Construct an optimization model with the objective of minimizing the overall electricity purchase cost for the project's main responsible unit. Decision variables include: the electricity purchase plan submitted to the main power grid, the energy storage charging and discharging plan, the electricity sales price for each internal user, and the green electricity allocation ratio. The electricity sales price for each internal user is differentiated based on their load characteristics and green electricity consumption contribution rate; the green electricity allocation ratio is determined according to the green electricity consumption priority ranking of each user.

[0012] The constraints of the optimization model include: power balance constraints (the power purchased by the main grid plus the output of photovoltaic / wind power plus the energy storage discharge power equals the sum of the loads of each user plus the energy storage charging power); energy storage constraints (upper and lower limits of state of charge, upper and lower limits of charging and discharging power, and cycle life constraints); green electricity consumption constraints (the proportion of green electricity of internal users is not lower than the preset threshold); interaction capacity constraints with the main grid (the purchased power does not exceed the declared capacity); and electricity sales price constraints (the electricity sales price of each user is between the electricity purchase price of the main grid and the price of electricity purchased directly from the grid by the user).

[0013] Preferably, the optimization model is solved using mixed integer linear programming with a time resolution of 15 minutes or 1 hour and a rolling optimization cycle of 24 hours.

[0014] Step S5: Solve the optimization model, output the power purchase strategy, energy storage scheduling strategy, electricity sales price and green electricity quota of each user for the project's main responsible unit in the future preset time period, and push the results to the edge computing device for execution.

[0015] Step S6 (Rolling Correction): Every 15 minutes, based on the latest measured load, photovoltaic output and electricity price, the power purchase plan, energy storage dispatch strategy, electricity sales price for each user and green electricity quota for the subsequent period are rolled over and corrected.

[0016] Secondly, this invention provides a multi-user load forecasting and power purchase decision-making system for the main responsible unit of a green electricity direct connection project, comprising:

[0017] The system includes a data acquisition module, a load forecasting module, a market information interface, an optimization decision-making module, an instruction issuance module, and a human-computer interaction interface.

[0018] The load forecasting module has a built-in classification model library, including high-fluctuation load forecasting models for industrial enterprises, periodic load forecasting models for commercial office buildings, and stable load forecasting models for data centers.

[0019] The optimization model constructed by the optimization decision module takes the minimum overall electricity purchase cost as the objective function. The constraints include power balance constraints, energy storage state of charge constraints, green electricity consumption constraints, interaction capacity constraints with the large power grid, and electricity sales price constraints.

[0020] The optimization decision-making module is also used to participate in the demand-side response market, automatically adjusting the power purchase plan and energy storage charging and discharging strategy according to the response instructions issued by the power grid company.

[0021] The optimization decision-making module is also used to continuously revise the power purchase plan, energy storage dispatch strategy, electricity sales price of each user, and green electricity quota every 15 minutes based on the latest measured load, photovoltaic output, and electricity price.

[0022] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0023] Compared with the prior art, the present invention has the following beneficial effects.

[0024] (1) For the first time, a dedicated tool is provided for the "project responsible unit": This fills the gap in traditional load forecasting and power purchase optimization methods in multi-user heterogeneous scenarios. By classifying and modeling by industry type, it adapts to the diversity of user load characteristics within the project responsible unit, and the forecasting accuracy is significantly improved.

[0025] (2) Multi-dimensional comprehensive optimization to reduce electricity purchase cost: The power purchase of the large power grid, internal photovoltaic / wind power consumption, energy storage dispatch, multi-user green electricity allocation and electricity sales pricing are optimized in a unified manner. By solving the decision scheme with the minimum overall electricity purchase cost through mixed integer linear programming, the overall electricity purchase cost of the project's main responsible unit can be reduced by 10%-20%.

[0026] (3) Differentiated electricity pricing and green electricity allocation: Based on the load characteristics of each user and the contribution rate of green electricity consumption, the electricity price and green electricity quota are dynamically determined, which not only incentivizes users to optimize their electricity consumption behavior, but also ensures that the overall green electricity consumption rate meets the policy requirements.

[0027] (4) Rolling correction mechanism adapts to real-time fluctuations: The subsequent plan is rolled and corrected every 15 minutes based on the latest measured data, so that the power purchase strategy can adapt to the real-time fluctuations of load and new energy output, and significantly improve the robustness and practicality of the strategy.

[0028] (5) Support the demand-side response market: The optimized decision-making module can receive demand-side response instructions issued by the power grid company, automatically adjust the power purchase plan and energy storage charging and discharging strategy, and further broaden the revenue sources of the project's main responsible unit. Attached Figure Description

[0030] Figure 1 This is a diagram showing the overall architecture of the system of the present invention.

[0031] Figure 2 This is a flowchart illustrating the workflow of the load forecasting module of the present invention.

[0032] Figure 3 This is a flowchart of the algorithm for the optimization decision module of this invention. Detailed Implementation

[0034] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0035] Example 1: System Architecture

[0036] like Figure 1 As shown, this embodiment provides a multi-user load forecasting and power purchase decision system for the main responsible unit of green electricity direct connection projects, including a data acquisition module (1), a load forecasting module (2), a market information interface (3), an optimization decision module (4), an instruction issuance module (5), and a human-computer interaction interface (6).

[0037] The data acquisition module (1) connects to each user side through smart meters and the user production management system to collect historical electricity consumption data (hourly power data for at least the past 12 months), industry type (industrial enterprises / commercial offices / data centers), production plans (such as factory work schedules, data center IT equipment expansion plans), and local meteorological data (temperature, humidity, light intensity). The data acquisition cycle is 15 minutes.

[0038] The load forecasting module (2) has a built-in classification model library, including: an LSTM high-fluctuation load forecasting model for industrial enterprises (input features include historical power, production shift indicators, temperature and humidity), a TCN periodic load forecasting model for commercial office buildings (input features include historical power, work and rest period indicators, and holiday indicators), and a regression model for data centers (input features include IT equipment load rate, cooling system power, and historical power). Each model combines quantile regression methods to output a load forecast interval with a 95% confidence interval.

[0039] The market information interface (3) connects to the power trading center platform via API to obtain the time-of-use electricity price (RMB / kWh) and green electricity premium (RMB / kWh) of the large power grid; obtains the predicted output curve for the next 24 hours through the local photovoltaic / wind power monitoring system; and obtains the current state of charge (SOC, %), rated charging and discharging power (MW) and cost per kilowatt-hour (RMB / kWh) through the energy storage EMS.

[0040] The optimization decision module (4) constructs a mixed-integer linear programming model with a time granularity of 15 minutes and an optimization window of the next 24 hours (96 time periods). The objective function is to minimize the total operating cost, including the power purchase cost of the main grid, the depreciation cost of energy storage, and the revenue from selling electricity to users. The decision variables are the power purchase capacity of the main grid, the charging and discharging power of energy storage, the electricity sales price of each user, and the green electricity allocation ratio in each time period. The constraints include: system power balance constraints, energy storage SOC maintained at 20%-90%, the green electricity ratio of each user not less than 50%, the power purchase capacity not exceeding the declared capacity, and the electricity sales price of each user between the power purchase price of the main grid and the price of electricity purchased directly from the grid by the user.

[0041] The instruction delivery module (5) sends the optimized electricity purchase plan to the edge computing device via MQTT or Modbus TCP protocol. The edge computing device then sends the energy storage charging and discharging instructions to the energy storage PCS and sends the green electricity quota and electricity sales price to the energy management system of each user.

[0042] The human-computer interaction interface (6) displays the total load forecast curve, user load forecast curve, power purchase plan bar chart, energy storage charging and discharging plan, green electricity quota and electricity sales price of each user, as well as expected total cost and expected revenue in real time in the form of a dashboard for the next 24 hours.

[0043] The optimization decision module (4) adjusts the power purchase plan, energy storage dispatch strategy, electricity sales price and green electricity quota for subsequent periods every 15 minutes based on the latest measured load, photovoltaic output and electricity price.

[0044] When the optimization decision module (4) receives a demand-side response instruction (such as requiring a reduction in power purchase within the next 2 hours) issued by the power grid company, the module adds the instruction as a mandatory constraint to the optimization model, solves the problem again, and automatically adjusts the energy storage discharge power and user-side load reduction suggestions.

[0045] Example 2: Load Forecasting Method

[0046] like Figure 2As shown, the workflow of the load forecasting module is as follows: Start → Receive user data pushed by the data acquisition module → Read user industry type tags → Select the corresponding forecasting model according to the industry type (industrial enterprises → high volatility LSTM model, commercial offices → periodic TCN model, data centers → stable load regression model) → Load the historical data of the last 30 days as model input → Model outputs the load forecast value for the next 24 hours every 15 minutes and the upper and lower limits of the 95% confidence interval → Push the forecast results to the optimization decision module → End.

[0047] For newly connected users, if the historical data is less than 30 days, the module will first use a general basic model of the same industry type for prediction. After accumulating enough data, it will automatically switch to the user's exclusive model.

[0048] Example 3: Optimizing the Decision-Making Process

[0049] like Figure 3 As shown, the workflow of the optimization decision module is as follows: Start → Triggered every 15 minutes or hour, or triggered by a rolling correction event → Obtain the load forecast values ​​for each user for the next 24 hours from the load forecast module → Obtain the latest time-of-use electricity price, green electricity premium, photovoltaic / wind power forecast output, and energy storage SOC from the market information interface → Construct a mixed integer linear programming model with the goal of minimizing total cost → Add power balance constraints, energy storage constraints, green electricity consumption constraints, interactive capacity constraints, and electricity sales price constraints → Call the solver to solve → Determine if there is a feasible solution → If there is a feasible solution, output the optimal electricity purchase plan, energy storage charging and discharging plan, electricity sales price for each user, and green electricity quota, and send them to the instruction issuance module; if there is no feasible solution, relax the constraints in order of priority from low to high (relax the green electricity consumption constraint first, then relax the electricity sales price constraint), and solve again → Push the optimization results to the human-computer interaction interface for display → Enter the next waiting period.

[0050] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for multi-user load forecasting and power purchase decision-making for the main responsible unit of a green electricity direct connection project, characterized in that, Includes the following steps: Step S1: Collect industry type, historical electricity consumption data, production plans, and meteorological data of each user within the project's responsible unit; Step S2: Classify users based on the industry type, establish a load forecasting model for each user category, and output the load forecast value for the future preset time period; Step S3: Obtain the time-of-use electricity price of the main power grid, the green electricity premium, the internal photovoltaic / wind power forecast output, the status of the energy storage system, and the cost per kilowatt-hour; Step S4: Construct an optimization model with the goal of minimizing the overall electricity purchase cost of the project's main responsible unit. Decision variables include the electricity purchase plan submitted to the main power grid, the energy storage charging and discharging plan, the electricity sales price of each internal user, and the green electricity allocation ratio. Step S5: Solve the optimization model, output the power purchase strategy, energy storage dispatch strategy, electricity sales price and green electricity quota of each user for the project's main responsible unit in the future preset time period, and push the results to the edge computing device for execution; Step S6: Every 15 minutes, based on the latest measured load, photovoltaic output and electricity price, continuously revise the power purchase plan, energy storage dispatch strategy, electricity sales price for each user and green electricity quota for subsequent periods.

2. The method according to claim 1, characterized in that, The load forecasting model described in step S2 uses a long short-term memory network or a temporal convolutional network, and combines it with a quantile regression method to output a load forecasting interval for assessing forecast uncertainty.

3. The method according to claim 1, characterized in that, In step S4, the electricity sales price for each internal user is differentiated based on the load characteristics and green electricity consumption contribution rate of each user; the green electricity allocation ratio is determined according to the green electricity consumption priority ranking of each user.

4. The method according to claim 1, characterized in that, The constraints of the optimization model described in step S4 include: power balance constraints, upper and lower limits of energy storage state of charge and charging and discharging power, green electricity consumption constraints where the proportion of green electricity in internal users is not lower than a preset threshold, interactive capacity constraints where the purchased power does not exceed the declared capacity, and electricity sales prices for each user that are between the electricity purchase price from the large power grid and the electricity purchase price from the grid.

5. The method according to claim 1, characterized in that, The optimization model is solved using mixed integer linear programming with a time resolution of 15 minutes or 1 hour and a rolling optimization cycle of 24 hours.

6. A multi-user load forecasting and power purchase decision-making system for the main responsible unit of a green electricity direct connection project, characterized in that, include: The data acquisition module is used to obtain information on the industry type, historical electricity consumption data, production plans, and meteorological data of each user within the project's responsible unit. The load forecasting module, connected to the data acquisition module, is used to output the load forecast values ​​for each user in the future preset time period based on the data using a categorical load forecasting model; the categorical load forecasting model includes a high-fluctuation load forecasting model for industrial enterprises, a periodic load forecasting model for commercial office buildings, and a stable load forecasting model for data centers. Market information interface, used to obtain time-of-use electricity prices of the main power grid, green electricity premium, internal photovoltaic / wind power forecast output, energy storage system status and cost per kilowatt-hour; The optimization decision module is connected to both the load forecasting module and the market information interface. It is used to construct an optimization model with the goal of minimizing the overall electricity purchase cost, and to solve for the electricity purchase strategy, energy storage dispatch strategy, electricity sales price for each user, and green electricity quota. The optimization model is solved using mixed integer linear programming with a time resolution of 15 minutes and a rolling optimization cycle of 24 hours. The constraints include power balance constraints, energy storage state of charge constraints, green electricity consumption constraints, interaction capacity constraints with the main power grid, and electricity sales price constraints. The instruction issuing module is connected to the optimization decision module and is used to push the optimization results to the edge computing device and each user side for execution. The human-computer interaction interface, connected to the optimization decision module, is used to display load forecast curves, power purchase plans, energy storage charging and discharging plans, and expected costs.

7. The system according to claim 6, characterized in that, The optimization decision-making module is also used to participate in the demand-side response market, automatically adjusting the power purchase plan and energy storage charging and discharging strategy according to the response instructions issued by the power grid company.

8. The system according to claim 6, characterized in that, The optimization decision-making module is also used to continuously revise the power purchase plan, energy storage dispatch strategy, electricity sales price of each user, and green electricity quota every 15 minutes based on the latest measured load, photovoltaic output, and electricity price for subsequent periods.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 5.

Images