A load optimization scheduling method for virtual power plant considering peak shaving potential of high-load users

Abstract

This invention discloses a virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users. The steps are as follows: Step 1: Collect multi-source data and perform preprocessing; Step 2: Construct an evaluation index system for the peak-shaving potential of high-energy-consuming users; Step 3: Quantify and combine the indexes with weights; Step 4: Map the user's peak-shaving potential to scheduling parameters; Step 5: Construct a virtual power plant optimization scheduling model; Step 6: Solve the model and output the results. This invention belongs to the field of virtual power plant scheduling technology in power systems, and solves the problems in existing technologies such as the difficulty in quantifying the peak-shaving potential of high-energy-consuming industrial users, the difficulty in incorporating potential evaluation results into the scheduling model, insufficient user-differentiated dispatch, and the difficulty in balancing aggregator revenue and user revenue.

Description

Technical Field

[0001] This invention belongs to the field of power system virtual power plant dispatching technology and is used for power system demand response, virtual power plant optimized operation and load-side resource dispatching management. It involves a virtual power plant load optimization dispatching method that considers the peak-shaving potential of high-energy-consuming users. Background Technology

[0002] With the continuous increase in the installed capacity of new energy sources such as wind power and photovoltaics, the randomness, volatility, and anti-peak-shaving characteristics of power system operation have become more prominent. Traditional methods relying on generation-side units for regulation are no longer sufficient to meet the flexible regulation needs under conditions of high-proportion new energy integration. Virtual power plants can aggregate resources such as distributed power sources, energy storage systems, and adjustable loads to participate in peak-shaving ancillary services and electricity market operation as a whole, which is an important way to improve the level of new energy consumption and system regulation capabilities. Among various load-side resources, high-energy-consuming industrial users have characteristics such as large load capacity, strong production continuity, high electricity cost ratio, and the ability to plan and adjust some aspects, providing a realistic basis for participating in system peak-shaving. Typical high-energy-consuming industries such as electrolytic aluminum, cement, and ferroalloys have significant differences in process flow, equipment start-up and shutdown constraints, inventory conditions, response speed, and management organization capabilities. Their adjustable capacity, duration, response reliability, and economic acceptability are not the same. Therefore, high-energy-consuming industrial users cannot be simply regarded as homogeneous loads for unified modeling and utilization.

[0003] Existing virtual power plant load dispatching methods still have several shortcomings in engineering applications: First, the assessment of user peak-shaving potential and the optimization dispatching process are disconnected, and the assessment results often remain at the level of resource ranking, capacity reporting, or post-event evaluation, failing to be directly transformed into constraint and incentive parameters in the dispatching model; Second, existing dispatching models usually adopt a uniform adjustable ratio, uniform response rate, or uniform subsidy price, making it difficult to reflect the differences in regulation depth, continuous response capability, and execution reliability among high-energy-consuming users; Third, some existing methods only focus on the amount of peak-shaving tasks completed or aggregator revenue, ignoring the changes in net energy costs after users participate in peak shaving, which can easily lead to insufficient user participation; Fourth, the synergistic effect between energy storage and industrial loads has not been fully characterized, and there is a lack of reasonable division of labor between the short-term compensation role of energy storage and the continuous regulation role of industrial loads.

[0004] Based on the aforementioned shortcomings, there is an urgent need to develop a virtual power plant load optimization scheduling method that can integrate "user potential identification - parameter mapping - optimized scheduling" to enable high-potential industrial users to be prioritized and rationally dispatched during critical peak-shaving periods. At the same time, differentiated subsidies and individual rational constraints should be used to ensure user participation benefits, thereby improving the completion rate and operational economy of the virtual power plant's two-way peak-shaving tasks. Summary of the Invention

[0005] The purpose of this invention is to provide a virtual power plant load optimization scheduling method that considers the peak-shaving potential of high-energy-consuming users, which solves the problems in the prior art such as the difficulty in quantifying the peak-shaving potential of high-energy-consuming industrial users, the difficulty in incorporating the potential assessment results into the scheduling model, the lack of differentiated user calls, and the difficulty in balancing the revenue of aggregators and the revenue of users.

[0006] The technical solution adopted in this invention is a virtual power plant load optimization scheduling method that considers the peak-shaving potential of high-energy-consuming users, implemented according to the following steps: Step 1: Collect and preprocess virtual power plant operation data, including historical load of high energy-consuming industrial users, peak shaving execution records, equipment operation information, user-declared adjustable power, new energy output forecast, conventional power output, energy storage parameters, time-of-use electricity price, and peak shaving demand issued by the grid.

[0007] Step 2: Construct an evaluation index system for the peak-shaving potential of high-energy-consuming users. The evaluation index system includes data-based hard indicators and evaluation-based soft indicators, and characterizes the peak-shaving potential of users from three dimensions: technology, economy and management.

[0008] Step 3: Quantify, standardize, and combine weights for the evaluation indicators. Use triangular fuzzy numbers to quantify soft indicators, use the FAHP method to determine subjective weights, use the CRITIC method to determine objective weights, and use game theory to combine weights to obtain comprehensive weights. Calculate the peak-shaving potential score for each user.

[0009] Step 4: Map user peak-shaving potential scores to scheduling model parameters, including reliable allocation upper limit, continuous response upper limit, and differentiated subsidy price, so that the potential scores are respectively incorporated into power constraints, time-series response constraints, and economic incentives.

[0010] Step 5: Establish a virtual power plant optimization scheduling model that considers the peak-shaving potential of users. With the goal of maximizing the net revenue of virtual power plants, the model comprehensively considers peak-shaving service revenue, user subsidy expenditure, energy storage operation cost, and peak-shaving task shortfall penalty cost. It also sets constraints such as industrial user load reconfiguration, reliable allocation upper limit constraint, continuous response upper limit constraint, daily power balance constraint, energy storage operation constraint, peak-shaving task tracking constraint, system power balance constraint, and individual rationality constraint.

[0011] Step 6: Solve the optimization scheduling model based on the valley filling and peak shaving service demands issued by the power grid, and output the power increase or decrease, energy storage charging and discharging plans, task deficits, net revenue of virtual power plants, and net energy cost of users for each industrial user in different time periods.

[0012] The beneficial effects of the present invention include the following aspects: 1) Achieve direct integration between potential assessment and scheduling models. By mapping user peak-shaving potential scores to credible allocation caps, continuous response caps, and differentiated subsidy prices, the assessment results are prevented from merely remaining at the ranking level, allowing user differences to be reflected simultaneously in scheduling constraints and economic incentives.

[0013] 2) Improve the rationality of dispatching power to high-energy-consuming industrial users. Taking into account load scale, maximum peak load, peak-to-valley ratio, load fluctuation rate, equipment start-up and shutdown rate, electricity price sensitivity coefficient, user response loss, system operation and shutdown switching difficulty, and system management coordination, it can simultaneously reflect the user's physical regulation capability, economic response capability, and organizational execution capability.

[0014] 3) Improve the completion rate of two-way peak shaving tasks in virtual power plants. By having industrial users and energy storage systems collaborate in valley filling and peak shaving services, high-potential users undertake core regulation tasks, while energy storage systems undertake short-term compensation and energy transfer tasks, thereby improving the accuracy of service windows in tracking peak shaving demands.

[0015] 4) Balancing aggregator revenue and user participation: The model aims to maximize the net revenue of the virtual power plant while setting individual user rationality constraints to ensure that the net energy cost after user participation in peak shaving is not higher than the baseline scenario, thereby enhancing the feasibility of the peak shaving scheme and the long-term stability of participation.

[0016] 5) Applicable to various types of high-energy-consuming industrial users. Based on the actual load curves and process constraints of different regions, industries, and enterprises, the system updates index data and model parameters, making it suitable for identifying peak-shaving potential and making virtual power plant dispatch decisions for high-energy-consuming users such as electrolytic aluminum, cement, and ferroalloys. Attached Figure Description

[0017] Figure 1 This is a flowchart of the peak-shaving potential assessment for high-energy-consuming users provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of user peak-shaving potential mapping to scheduling parameters and scheduling embedding provided in an embodiment of the present invention. Detailed Implementation

[0018] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments.

[0019] The method of this invention quantifies the peak-shaving potential of high-energy-consuming users and maps the evaluation results to the reliable allocation upper limit, continuous response upper limit and differentiated subsidy price in the scheduling model, thereby realizing the coordinated peak-shaving of industrial load and energy storage resources.

[0020] Reference Figure 1 and Figure 2 The overall technical solution of the method of the present invention is divided into six core steps, of which steps 2 and 3 correspond to Figure 1The high-energy-consuming user peak-shaving potential evaluation process shown in the figure corresponds to steps 4 and 5. Figure 2 The process of mapping user peak-shaving potential to scheduling parameters and embedding scheduling, as shown in the diagram, is implemented according to the following steps: Step 1: Collect multi-source data and perform preprocessing. 1.1) Collect multi-source data, including: Collect historical daily load curves, enterprise production plans, key equipment operating status, historical peak-shaving execution records, and parameters of adjustable and de-adjustable power declared by users for high-energy-consuming industrial users within the virtual power plant. Collect parameters such as new energy output forecast, small conventional power output, other loads, rated power of energy storage, rated capacity of energy storage, charge and discharge efficiency, upper / lower limits of state of charge, initial state of charge, and final state of charge. Collect initial values ​​of time-of-use electricity pricing, grid peak-shaving service prices, user subsidy prices, and deficit penalty coefficients.

[0021] 1.2) Perform outlier identification, missing value imputation, time scale unification, and indicator standardization on the collected multi-source data; If the scheduling cycle is 24 hours and the time interval is 1 hour, then all types of time series data will be unified into 24 time periods; if a 15-minute interval is used, then it will be unified into 96 time periods.

[0022] Step 2: Construct an evaluation index system for the peak-shaving potential of high-energy-consuming users. The evaluation index system for the peak-shaving potential of high-energy-consuming users consists of technical, economic and management dimensions, including four data-based hard indicators and five evaluation-based soft indicators; 2.1) Data-driven hard indicators are calculated from users' historical load curves and peak shaving execution records. Data-driven hard indicators include average load C1, maximum peak load C2, peak-to-valley ratio C3, and load fluctuation rate C4. The four data-driven hard indicators are explained in detail below: Average load C1: The arithmetic mean of load data over the entire period, used to assess the overall electricity consumption level of users. Combined with the peak-valley difference parameter, it can comprehensively analyze the load fluctuation characteristics. The expression is: (1) In the formula, N The value is 96 (dividing the day into 15-minute intervals). L t Indicates the first t Load observations for each time period; Maximum peak load C2: Used to assess the changeability and transferability of the load's electricity consumption behavior, expressed as: (2) In the formula, L max This represents the peak daily load for users. L avg This represents the average load for users during periods when they are not participating in peak shaving. Peak-to-valley ratio C3: Reflects the degree of difference between peak and valley loads, expressed as: (3) In the formula, L pp This represents the load value during peak electricity consumption periods for users. L v This refers to the load value during off-peak electricity consumption periods for users. Load volatility C4: The ratio of the standard deviation of user load to the mean during peak shaving periods, quantifying the relative intensity of load volatility, expressed as: (4) (5) In the formula, Let be the standard deviation of the load for user i; L cavg This represents the average load during the peak-shaving period for this user. m and M Do not specify the start and end time periods for peak-shaving service periods; For users i During peak service hours t The load value; when the load fluctuation rate is large, the load of industrial users changes drastically and is unstable, making it more difficult to predict and manage peak load.

[0023] 2.2) Evaluation-type soft indicators are obtained by fuzzy quantification of expert evaluation, enterprise production organization data and historical response performance. The five evaluation-type soft indicators include equipment start-up and shutdown rate C5, electricity price sensitivity coefficient C6, user response loss C7, system operation and shutdown switching difficulty C8 and system management coordination degree C9.

[0024] Step 3: Indicator Quantification and Combined Weighting For soft indicators such as equipment start-up and shutdown rate (C5), electricity price sensitivity coefficient (C6), user response loss (C7), system operation and shutdown switching difficulty (C8), and system management coordination degree (C9), which are difficult to calculate directly using historical load curves, triangular fuzzy numbers (TFCs) are used. l,m,u Semantic quantization is performed, and the centroid method is used to obtain the sharpness value. The expression is as follows: (6) in, x The sharpness value after deblurring by the triangular blur number. l u kThese represent the lower limit, median, and upper limit of the triangular fuzzy number, respectively. Positive indicators are processed using benefit-based standardization, while negative indicators are processed using inverse standardization, resulting in a standardized evaluation matrix. The FAHP method is used to construct a fuzzy judgment matrix and calculate subjective weights. Then, the CRITIC method is used to calculate the objective weights based on the standard deviation of the indicators and the correlation between the indicators. Then, the overall weight is determined by combining weights using game theory. The overall weight and user peak-shaving potential score satisfy the following relationships: (7) (8) in, i Number the user; j The evaluation indicator number; w j For the first j The overall weight of each evaluation indicator; S i For users i Peak-shaving potential score; z ij For users i In the j Standardized values ​​under each evaluation indicator; n This represents the total number of evaluation indicators.

[0025] Thus, based on the user's peak-shaving potential score, high-energy-consuming industrial users are sorted and classified into high, medium, and low peak-shaving potential levels, providing parameter basis for subsequent differentiated scheduling.

[0026] Step 4: Map user peak-shaving potential to scheduling parameters. To embed the evaluation results into the virtual power plant optimal scheduling model, the user is first... i The peak-shaving potential score is normalized into a potential coefficient. The expression is: (9) In the formula, the lowest potential score and highest potential rating Taken from the set of users participating in the scheduling respectively; potential coefficient The larger the value, the higher the user's peak-shaving potential; Regarding power allocation, the user-reported adjustable power will be corrected to a trusted allocation limit, expressed as follows: (10) In the formula, For users i During the period tThe upper limit of trusted allocation; For users i During the period t The declared adjustable power; and Assign upper bound mapping coefficients to trusted entities; For users i Potential coefficient; Regarding time-series response, the user potential coefficient is mapped to the upper limit of continuous response, expressed as: (11) In the formula, Let be the upper limit of continuous responses for user i; The upper limit of the basic continuous response; For response adjustment coefficient; In terms of economic incentives, the user potential coefficient is mapped to the valley filling subsidy price and the peak shaving subsidy price, expressed as follows: (12) (13) In the formula, and users respectively i Unit subsidy price when participating in valley filling and peak shaving services; and These are the basic subsidy prices for valley filling services and peak shaving services, respectively. γ This is the price adjustment coefficient.

[0027] Generally, the system value of peak shaving services is higher than that of valley filling services. Therefore, the base subsidy price for peak shaving is higher than that for valley filling. Through the above mapping, high-potential users can obtain a larger trusted allocation limit, a higher continuous response limit, and a higher unit subsidy.

[0028] Step 5: Construct a virtual power plant optimization scheduling model. Let the set of scheduling periods be . T The set of high-energy-consuming users participating in the scheduling is I The power grid during the period t Issue peak-shaving demand; when the demand is positive, it represents valley-filling demand, and when the demand is negative, it represents peak-shaving demand. Valley-filling demand and peak-shaving demand are defined separately, with the following expressions: (14) (15) In the formula, For the power grid during the time period t The issued peak-shaving requirements; For time period t The demand for filling the valley; For time period t Peak shaving demand; and when When positive, it indicates the demand for filling the valley. A negative value indicates peak shaving demand; With the goal of maximizing the net profit of the virtual power plant, the following objective function is established, expressed as: (16) In the formula, Net revenue from virtual power plants; , , , Let $\frac{ ... (17) (18) (19) (20) In the formula, Ω up and Ω down These are the valley filling service period set and the peak shaving service period set, respectively. and points Do not pay the unit service price for the power grid for valley filling and peak shaving services; e t up and e t down Time periods t Task shortages in the valley filling and peak shaving directions; In the formula, r i up and r i down These are the unit subsidy prices for user i when participating in valley filling and peak shaving services, respectively. x i,t up and x i,t down Let represent the power increase and power decrease for user i during time period t, respectively; I represents the set of high-energy-consuming users participating in the scheduling. In the formula, p t ch and pt dis These represent the charging power and discharging power of the energy storage during time period t, respectively. c es The charging and discharging cost of energy storage units; T For the set of scheduling periods; In the formula, Cmiss The cost of penalties for shortfall; The penalty coefficient per unit of shortfall; and Time periods t There is a shortage of tasks in the valley filling and peak reduction directions.

[0029] Among the parameters mentioned above, the service price, task deficit, user power adjustment / decrease, energy storage charging and discharging power, energy storage unit operating cost, and unit deficit penalty coefficient are all taken according to their corresponding peak shaving direction and time period. The virtual power plant optimization scheduling model includes at least the following constraints: (1) User load reconfiguration constraints: The load after dispatch is determined by the baseline load, the upward adjustment power, and the downward adjustment power, and its expression is: (twenty one) in, For users i During the period t The actual load after scheduling; For users i During the period t The baseline load; For users i During the period t Increased power; For users i During the period t The power reduction; (2) Trusted Allocation Upper Limit Constraint: The power increase and decrease of a user shall not exceed its trusted allocation upper limit, the expression of which is: (twenty two) in, For users i During the period t The upper limit of trusted allocation; and users respectively i During the period t The upward and downward adjustment of power; (3) Continuous response upper limit constraint: The change in net regulation power between adjacent time periods shall not exceed the user's continuous response upper limit, the expression of which is: (twenty three) in, For users i The upper limit of continuous response; and Representing users respectively i During the period t Net regulating power; and Representing users respectively i Net regulation power in the previous period; (4) Daily power balance constraint: The deviation of the total daily power consumption before and after peak shaving shall not exceed the allowable range, and its expression is: (twenty four) in, For users i Allowable daily power consumption deviation within the scheduling cycle; T For the set of scheduling periods; and users respectively i During the period t The upward and downward adjustment of power; (5) Energy storage power constraint: Both the charging power and discharging power of energy storage must meet the rated power boundary, the expression of which is: (25) (26) in, and These represent the charging power and discharging power of the energy storage during time period t, respectively. and These are the maximum charging power and maximum discharging power of the energy storage, respectively. (6) Energy storage state of charge constraint: The energy storage state of charge changes dynamically according to the charging and discharging power, and always meets the upper and lower limit constraints. Its expression is: (27) (28) in, and Energy storage time period t The state of charge in the previous period; and These refer to energy storage charging efficiency and discharging efficiency, respectively. and These are the lower and upper limits of the energy storage state of charge, respectively; This refers to the rated capacity of the energy storage. (7) Peak shaving task tracking constraints: During the valley filling period, the task is jointly undertaken by industrial users increasing their load and energy storage charging; during the peak shaving period, the task is jointly undertaken by industrial users decreasing their load and energy storage discharging. The expression is as follows: (29) (30) in, and Time periods t Valley filling and peak shaving demand; and These are the task vacancies in the corresponding directions; and These are the energy storage charging power and the discharging power, respectively. and These options allow users to increase or decrease power, respectively. (8) System power balance constraint: Purchased electricity is used as a system balancing item to compensate for the net power difference after internal resource adjustment. Its expression is: (31) in, The purchased power during time period t; For time period t Other loads; For time period t The output of new energy sources; For time period t Small conventional power supply output; For users i During the period t The actual load after scheduling; and These are the energy storage charging power and the discharging power, respectively. (9) Individual rationality constraint: The net energy cost of users participating in peak shaving is not higher than that of the baseline scenario, and its expression is: (32) in, For users i Net energy cost in scheduling scenarios; For users i Net energy cost in the baseline scenario.

[0030] Step 6: Complete the model solution and output the results. The objective function and main constraints mentioned above can all be expressed in linear form, so a linear programming solver can be used to solve them; when further introducing mutually exclusive states of energy storage charging and discharging or user start-up and shutdown states, a mixed-integer linear programming solver can be used to solve them. After the solution is completed, the output includes the time-of-use power adjustment for each user, the energy storage charging and discharging power and SOC trajectory, the service window task deficit, the virtual power plant service revenue, the user subsidy expenditure, the energy storage operating cost, the virtual power plant net revenue, and the net energy cost for each user. Furthermore, this information can be displayed to users visually.

[0031] In the following embodiments, sunny, cloudy, and rainy days are used to characterize three typical operating conditions: high, medium, and low predicted output of renewable energy, respectively. Different weather conditions affect the predicted output of renewable energy and further alter the peak-shaving demand issued by the power grid. To verify the applicability of the method of the present invention under different renewable energy output conditions and different energy storage capacities, two types of energy storage parameters are set: energy storage configuration A and energy storage configuration B. Energy storage configuration A is a basic energy storage configuration with a rated capacity of 80,000 kWh; energy storage configuration B is an expanded energy storage configuration with a rated capacity of 120,000 kWh. The maximum charging power and maximum discharging power of both types of energy storage configurations are 10,000 kW, the charging efficiency and discharging efficiency are both 0.93, the lower limit of SOC is 0.15, the upper limit of SOC is 0.90, and the initial SOC and terminal SOC are both 0.45.

[0032] Example 1 This Example 1 illustrates a virtual power plant bidirectional peak-shaving implementation under clear weather conditions. Example 1 uses energy storage configuration A to demonstrate the implementation process of the method under conditions of high photovoltaic output, significant midday valley filling demand, and evening peak-shaving demand. The scheduling cycle is 24 hours, with a 1-hour time interval. High-energy-consuming industrial users participating in peak shaving are a, b, f, c, and g, covering high, medium, and low peak-shaving potential levels. The output of new energy sources and small conventional power sources is used as a given quantity, and purchased electricity is used as a system power balance term. Industrial users and energy storage jointly undertake the bidirectional peak-shaving task assigned by the grid.

[0033] Step 1: Virtual power plant operation data acquisition and preprocessing. Historical load data, peak-shaving execution records, user-declared adjustable power, renewable energy output forecasts, conventional power output, energy storage operating parameters, time-of-use pricing, and grid-issued peak-shaving demands are collected from high-energy-consuming industrial users a, b, f, c, and g. In this Example 1, energy storage configuration A has a rated capacity of 80,000 kWh, a maximum charging power and a maximum discharging power of 10,000 kW, a charging efficiency and a discharging efficiency of 0.93, a lower limit of SOC of 0.15, an upper limit of SOC of 0.90, and an initial SOC and a final SOC of 0.45. The time-of-use pricing is 0.86 yuan / kWh for peak periods, 0.52 yuan / kWh for flat periods, and 0.28 yuan / kWh for valley periods. After imputing missing values, correcting outliers, and unifying the time scale of the collected data, user baseline load sequences, energy storage parameter sequences, renewable energy output sequences, and grid peak-shaving demand sequences are formed for 24 scheduling periods.

[0034] Step 2: Construction of an evaluation index system for the peak-shaving potential of high-energy-consuming users. The indicator system includes four data-driven hard indicators and five evaluation-driven soft indicators. The data-driven hard indicators include average load, maximum peak load, peak-to-valley ratio, and load fluctuation rate; the evaluation-driven soft indicators include equipment start-up and shutdown rate, electricity price sensitivity coefficient, user response loss, system operation and shutdown switching difficulty, and system management coordination.

[0035] Among them, the average load, maximum peak load, peak-to-valley ratio, and load fluctuation rate are calculated from the user's historical load data; the equipment start-up and shutdown rate, electricity price sensitivity coefficient, user response loss, system operation and shutdown switching difficulty, and system management coordination are obtained by fuzzy quantification based on expert evaluation, enterprise production organization data, and historical response performance.

[0036] Step 3: Quantify and weight the evaluation indicators. The evaluation indicators obtained in step 2 are quantified and standardized. For evaluation-type soft indicators, triangular fuzzy numbers are used for semantic quantification, and centroid method is used for defuzzification. For positive indicators, benefit-type standardization is used, and for negative indicators, inverse standardization is used. Subjective weights are calculated using the FAHP method, objective weights are calculated using the CRITIC method, and comprehensive weights are obtained through game theory combination weighting. In this embodiment 1, the comprehensive weights of average load, maximum peak load, load fluctuation rate, peak-to-valley ratio, equipment start-up and shutdown rate, electricity price sensitivity coefficient, user response loss, system operation and shutdown switching difficulty, and system management coordination are 0.159, 0.139, 0.098, 0.130, 0.092, 0.149, 0.084, 0.074, and 0.075, respectively.

[0037] Based on the standardized evaluation matrix and comprehensive weights, the peak-shaving potential scores for users a, b, f, c, and g were calculated to be 0.419, 0.330, 0.292, 0.287, and 0.283, respectively. These scores serve as input for subsequent scheduling parameter mapping.

[0038] Step 4: Map user peak-shaving potential scores to scheduling parameters. The user peak-shaving potential score is normalized into a user potential coefficient, and further mapped to a credible allocation cap, a continuous response cap, and a differentiated subsidy price.

[0039] Table 1 shows the trusted allocation coefficients, continuous response limits, and differentiated subsidy prices for valley filling and peak shaving periods for the five users a, b, f, c, and g.

[0040] Table 1. Mapping Results of Peak Shaving Potential to Scheduling Parameters for Each User

[0041] Therefore, users with higher peak-shaving potential have a larger reliable allocation ceiling, a higher continuous response ceiling, and a higher unit subsidy price in the scheduling model.

[0042] Step 5: Construct a virtual power plant optimization scheduling model considering user peak-shaving potential. Based on the bidirectional peak-shaving demand issued by the power grid, a virtual power plant optimal dispatch model is established that considers the peak-shaving potential of users. The model aims to maximize the net revenue of the virtual power plant, comprehensively considering peak-shaving service revenue, user subsidy expenditure, energy storage operating costs, and penalty costs for insufficient peak-shaving tasks.

[0043] In this Example 1, the power grid issues valley filling tasks from 11:00 to 14:00, with demands of 18MW, 20MW, 20MW, and 18MW respectively, for a total valley filling demand of 76MWh; and issues peak shaving tasks from 18:00 to 21:00, with demands of 16MW, 20MW, 20MW, and 16MW respectively, for a total peak shaving demand of 72MWh. The model includes constraints on user load reconfiguration, reliable allocation upper limit, continuous response upper limit, daily power balance, energy storage operation, peak shaving task tracking, system power balance, and individual user rationality.

[0044] Among them, the trusted allocation upper limit constraint is used to restrict the user's power increase and decrease from exceeding the trusted allocation upper limit corrected by the user's potential coefficient; the continuous response upper limit constraint is used to restrict the change of the user's net adjustment power in adjacent time periods; and the differentiated subsidy settlement relationship is used to calculate the user subsidy according to the user's potential coefficient and the actual adjustment power.

[0045] Step 6: Model Solving and Result Output The following scenarios were solved: S1 (comparison scenario) which did not consider differences in user peak-shaving potential, and S2 scenario which used the method of this invention. S1 scenario used uniform scheduling parameters, while S2 scenario used differentiated scheduling parameters obtained in step 4. Relevant data for this embodiment 1 are shown in Table 3.

[0046] The results show that: in scenario S1, 67.21 MWh of valley filling and 63.56 MWh of peak shaving were completed, for a total peak shaving amount of 130.77 MWh and a total deficit of 17.23 MWh, with a task completion rate of 88.36%; in scenario S2, 76.00 MWh of valley filling and 72.00 MWh of peak shaving were completed, for a total peak shaving amount of 148.00 MWh and a total deficit of 0, with a task completion rate of 100.00%.

[0047] Table 2. Comparison of Virtual Power Plant Revenue in Different Scenarios

[0048] Referring to Table 2, from an economic perspective: In scenario S1, the total revenue from virtual power plant services is RMB 154,919.09, the penalty for shortfall is RMB 20,671.50, the energy storage cost is RMB 1,038.38, the user subsidy is RMB 22,360.07, and the net profit of the virtual power plant is RMB 110,849.14; in scenario S2, the total revenue from virtual power plant services is RMB 175,440.00, the penalty for shortfall is 0, the energy storage cost is RMB 1,038.38, the user subsidy is RMB 52,311.85, and the net profit of the virtual power plant is RMB 122,089.77. Compared to scenario S1, the net profit of the virtual power plant in scenario S2 increases by RMB 11,240.63.

[0049] Meanwhile, the net energy costs for users a, b, f, c, and g in scenario S2 are RMB 416,038.75, RMB 347,361.05, RMB 255,546.25, RMB 293,811.42, and RMB 138,447.48, respectively, all lower than the corresponding baseline scenario and scenario S1. This demonstrates that the method of this invention can improve the completion rate and net revenue of virtual power plant peak-shaving tasks while reducing users' net energy costs and enhancing the enthusiasm of high-energy-consuming users to participate in peak-shaving.

[0050] Example 2 This Example 2 is an example of virtual power plant peak shaving dispatch under rainy conditions. Example 2 still uses energy storage configuration A, which has a rated capacity of 80,000 kWh, a maximum charging power and maximum discharging power of 10,000 kW, a charging efficiency and discharging efficiency of 0.93, a lower limit of SOC of 0.15, an upper limit of SOC of 0.90, and an initial SOC and terminal SOC of 0.45. User parameters remain the same as those shown in Table 1. Under rainy conditions, photovoltaic output is low, and the surplus of new energy is not significant during the midday period. Therefore, the grid does not issue valley filling tasks from 11:00 to 14:00. During the evening peak load period, the support capacity of new energy is weak, and the grid issues peak shaving tasks from 18:00 to 21:00, with demands of 18MW, 22MW, 22MW, and 18MW respectively, for a total peak shaving demand of 80MWh.

[0051] In this embodiment 2, two scenarios, S1 and S2, are also set up. S1 is a unified parameter scheduling scenario that does not consider the differences in user peak-shaving potential, and S2 is a peak-shaving potential differentiated scheduling scenario using the method of this invention. The relevant data for this embodiment 2 are shown in Table 3.

[0052] The results show that in scenario S1, 70.00 MWh of peak shaving was achieved, with a task deficit of 10.00 MWh, a completion rate of 87.50%, virtual power plant service revenue of RMB 148,400.00, and net profit of RMB 77,540.00. In scenario S2, 80.00 MWh of peak shaving was achieved, with no task deficit, a completion rate of 100.00%, virtual power plant service revenue of RMB 169,600.00, and net profit of RMB 91,154.50. Therefore, under the same rainy weather and energy storage configuration A, the method of this invention can improve the peak shaving task completion rate, reduce the task deficit, and increase the net profit of the virtual power plant.

[0053] Example 3 This embodiment 3 is a virtual power plant bidirectional peak shaving embodiment under cloudy conditions. This embodiment 3 still uses energy storage configuration A, which has a rated capacity of 80,000 kWh, a maximum charging power and a maximum discharging power of 10,000 kW, a charging efficiency and a discharging efficiency of 0.93, a lower limit of SOC of 0.15, an upper limit of SOC of 0.90, and an initial SOC and a terminal SOC of 0.45. User parameters are retained from the values ​​shown in Table 1. Under cloudy conditions, photovoltaic output is lower than on sunny days but still has some output at midday. Therefore, the grid issues a medium-intensity valley filling task from 11:00 to 14:00, with demands of 8MW, 10MW, 10MW, and 8MW respectively, for a total valley filling demand of 36MWh; and a peak shaving task from 18:00 to 21:00, with demands of 16MW, 22MW, 22MW, and 16MW respectively, for a total peak shaving demand of 76MWh. The total peak shaving task for the whole day is 112MWh.

[0054] In this embodiment 3, two scenarios, S1 and S2, are set up. S1 uses uniform scheduling parameters and does not consider the differences in user peak-shaving potential; S2 uses the method of this invention to map user peak-shaving potential scores to a reliable allocation upper limit, a continuous response upper limit, and a differentiated subsidy price. The relevant data for this embodiment 3 are shown in Table 3.

[0055] Table 3. Comparison of S1 / S2 scheduling results under Examples 1-3

[0056] The results show that in scenario S1, 32.00 MWh of valley filling and 67.50 MWh of peak shaving were completed, for a total peak shaving volume of 99.50 MWh, with a task deficit of 12.50 MWh and a task completion rate of 88.84%. The virtual power plant service revenue was RMB 152,700.00, and the net profit was RMB 106,360.00. In scenario S2, 36.00 MWh of valley filling and 76.00 MWh of peak shaving were completed, for a total peak shaving volume of 112.00 MWh, with no task deficit and a task completion rate of 100.00%. The virtual power plant service revenue was RMB 171,920.00, and the net profit was RMB 122,285.00. Therefore, under cloudy weather conditions, the method of this invention can simultaneously adapt to scheduling scenarios where valley filling demand decreases at midday and peak shaving demand increases in the evening, and improves the peak shaving performance and operational revenue of the virtual power plant under the same external conditions.

[0057] Example 4 This embodiment 4 is a virtual power plant bidirectional peak shaving embodiment under clear weather conditions. This embodiment 4 uses energy storage configuration B, which is an expanded capacity energy storage configuration with a rated capacity of 120,000 kWh. Compared with energy storage configuration A, energy storage configuration B only increases the rated capacity, while the maximum charging power and maximum discharging power remain at 10,000 kW, illustrating the implementation effect of the method of the present invention under conditions of increased available energy storage space. In this embodiment 4, the rated capacity of energy storage configuration B is 120,000 kWh, the maximum charging power and maximum discharging power are both 10,000 kW, the charging efficiency and discharging efficiency are both 0.93, the lower limit of SOC is 0.15, the upper limit of SOC is 0.90, and the initial SOC and terminal SOC are both 0.45. Except for the different energy storage capacity, the weather conditions, grid-issued peak shaving tasks, user aggregators, user peak shaving potential scoring, and parameter mapping methods in this embodiment 4 are consistent with those in embodiment 1. Under clear weather conditions, the power grid will issue valley filling tasks from 11:00 to 14:00, with demands of 18MW, 20MW, 20MW, and 18MW respectively, for a total valley filling demand of 76MWh; and will issue peak shaving tasks from 18:00 to 21:00, with demands of 16MW, 20MW, 20MW, and 16MW respectively, for a total peak shaving demand of 72MWh. The total peak shaving task for the entire day is 148MWh.

[0058] In this embodiment 4, two scenarios, S1 and S2, are set up. S1 is a unified parameter scheduling scenario that does not consider the differences in user peak-shaving potential, and S2 is a peak-shaving potential differentiated scheduling scenario using the method of this invention. The relevant data for this embodiment 4 are shown in Table 4.

[0059] The results show that in scenario S1, 70.50 MWh of valley filling and 67.00 MWh of peak shaving were completed, for a total peak shaving capacity of 137.50 MWh. The task deficit was 10.50 MWh, resulting in a task completion rate of 92.91%. Virtual power plant service revenue was RMB 163,190.00, and net profit was RMB 115,930.00. In scenario S2, 76.00 MWh of valley filling and 72.00 MWh of peak shaving were completed, for a total peak shaving capacity of 148.00 MWh. The task deficit was zero, resulting in a task completion rate of 100.00%. Virtual power plant service revenue was RMB 175,440.00, and net profit was RMB 123,645.00. Compared to energy storage configuration A, energy storage configuration B increases the available energy space for energy storage, allowing some regulation tasks to be undertaken by energy storage, thereby changing the composition of user subsidy expenditures and energy storage operating costs. Therefore, under the condition of the same service revenue, the net profit of the virtual power plant is improved. This demonstrates that, under clear weather conditions and energy storage configuration B, the method of this invention can still achieve full completion of peak-shaving tasks through differentiated user potential utilization, and further improve the net profit of the virtual power plant.

[0060] Example 5 This embodiment 5 is a virtual power plant peak-shaving dispatch embodiment under rainy weather conditions. This embodiment 5 uses energy storage configuration B, which has a rated capacity of 120,000 kWh, a maximum charging power and a maximum discharging power of 10,000 kW, a charging efficiency and a discharging efficiency of 0.93, a lower limit of SOC of 0.15, an upper limit of SOC of 0.90, and an initial SOC and a terminal SOC of 0.45. Except for the change in energy storage capacity, the weather conditions and peak-shaving tasks in this embodiment 5 are consistent with those in embodiment 2. Under rainy weather conditions, the grid does not issue valley-filling tasks from 11:00 to 14:00; and issues peak-shaving tasks from 18:00 to 21:00, with demands of 18MW, 22MW, 22MW, and 18MW respectively, for a total peak-shaving demand of 80MWh.

[0061] In this embodiment 5, two scenarios, S1 and S2, are set up. S1 uses uniform scheduling parameters and does not consider the differences in user peak-shaving potential; S2 uses the method of this invention to determine the reliable allocation upper limit, continuous response upper limit, and differentiated subsidy price based on the user peak-shaving potential score. The relevant data for this embodiment 5 are shown in Table 4.

[0062] The results show that in scenario S1, 74.00 MWh of peak shaving was achieved, with a task deficit of 6.00 MWh, a completion rate of 92.50%, virtual power plant service revenue of RMB 156,880.00, and net profit of RMB 89,520.00. In scenario S2, 80.00 MWh of peak shaving was achieved, with no task deficit, a completion rate of 100.00%, virtual power plant service revenue of RMB 169,600.00, and net profit of RMB 105,931.70. Therefore, under rainy weather conditions and energy storage configuration B, the method of this invention can fully utilize the synergistic effect of high-potential users and energy storage resources, improve the completion rate of peak shaving tasks at night, and increase the net profit of virtual power plants.

[0063] Example 6 This embodiment 6 is a virtual power plant bidirectional peak shaving embodiment under cloudy conditions. This embodiment 6 uses energy storage configuration B, which has a rated capacity of 120,000 kWh, a maximum charging power and a maximum discharging power of 10,000 kW, a charging efficiency and a discharging efficiency of 0.93, a lower limit of SOC of 0.15, an upper limit of SOC of 0.90, and an initial SOC and a terminal SOC of 0.45. Except for the change in energy storage capacity, the weather conditions and peak shaving tasks in this embodiment 6 are consistent with those in embodiment 3. Under cloudy conditions, the grid issues valley filling tasks from 11:00 to 14:00, with demands of 8MW, 10MW, 10MW, and 8MW respectively, for a total valley filling demand of 36MWh; and issues peak shaving tasks from 18:00 to 21:00, with demands of 16MW, 22MW, 22MW, and 16MW respectively, for a total peak shaving demand of 76MWh. The total peak shaving task for the entire day is 112MWh.

[0064] In this embodiment 6, two scenarios, S1 and S2, are set up. S1 is a unified parameter scheduling scenario that does not consider the differences in user peak-shaving potential, and S2 is a peak-shaving potential differentiated scheduling scenario using the method of this invention. The relevant data for this embodiment 6 are shown in Table 4.

[0065] The results show that in scenario S1, 33.50 MWh of valley filling and 71.00 MWh of peak shaving were completed, for a total peak shaving of 104.50 MWh, with a task deficit of 7.50 MWh and a task completion rate of 93.30%. The virtual power plant service revenue was RMB 160,570.00, and the net profit was RMB 113,850.00. In scenario S2, 36.00 MWh of valley filling and 76.00 MWh of peak shaving were completed, for a total peak shaving of 112.00 MWh, with no task deficit and a task completion rate of 100.00%. The virtual power plant service revenue was RMB 171,920.00, and the net profit was RMB 122,285.00. In multi-cloud scenarios, since the maximum charging and discharging power of energy storage remains constant, energy storage configuration A is sufficient to meet the charging and discharging needs within the main service window. Therefore, increasing the energy storage capacity does not significantly improve the net profit of scenario S2. Therefore, under cloudy weather conditions and energy storage configuration B, the method of the present invention can coordinate and allocate peak-shaving tasks according to the differences in users' peak-shaving potential and the operating status of energy storage, so that the virtual power plant can obtain higher operating benefits while completing all peak-shaving tasks.

[0066] Table 4. Comparison of S1 / S2 scheduling results under Examples 4-6

[0067] As shown in Tables 3 and 4, under the same weather conditions and energy storage configuration, the task completion rate in scenario S2 is higher than that in scenario S1, the task deficit is lower, and the net revenue of the virtual power plant is also higher. These results indicate that mapping user peak-shaving potential scores to a reliable allocation upper limit, a continuous response upper limit, and a differentiated subsidy price can improve the coordinated scheduling capability of virtual power plants for high-energy-consuming industrial users and energy storage resources, thereby enhancing peak-shaving task completion and operational economy.

Claims

1. A virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users, characterized in that, Follow these steps: Step 1: Collect data from multiple sources and perform preprocessing; Step 2: Construct an evaluation index system for the peak-shaving potential of high-energy-consuming users; Step 3: Indicator Quantification and Combined Weighting; Step 4: Map user peak-shaving potential to scheduling parameters; Step 5: Construct a virtual power plant optimization scheduling model; Step 6: Complete the model solution and output the results.

2. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 1, characterized in that, In step 1, the specific process is as follows: 1.1) Collect multi-source data, including: Collect historical daily load curves, enterprise production plans, key equipment operating status, historical peak-shaving execution records, and parameters of adjustable and de-adjustable power declared by users for high-energy-consuming industrial users within the virtual power plant. Collect parameters such as new energy output forecast, small conventional power output, other loads, rated power of energy storage, rated capacity of energy storage, charge and discharge efficiency, upper / lower limits of state of charge, initial state of charge, and final state of charge. Collect initial values ​​of time-of-use electricity pricing, grid peak-shaving service prices, user subsidy prices, and deficit penalty coefficients; 1.2) Perform outlier identification, missing value imputation, time scale unification, and indicator standardization on the collected multi-source data.

3. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 1, characterized in that, In step 2, the specific process is as follows: The evaluation index system for the peak-shaving potential of high-energy-consuming users includes four data-based hard indicators and five evaluation-based soft indicators; 2.1) Data-driven hard indicators are calculated from users' historical load curves and peak shaving execution records. The four data-driven hard indicators include average load, maximum peak shaving load, peak-to-valley ratio, and load fluctuation rate. 2.2) Evaluation-type soft indicators are obtained by fuzzy quantification of expert evaluation, enterprise production organization data and historical response performance. The five evaluation-type soft indicators specifically include equipment start-up and shutdown rate, electricity price sensitivity coefficient, user response loss, system operation and shutdown switching difficulty and system management coordination.

4. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 3, characterized in that, The four data-driven hard indicators are explained in detail below: Average load C1: The arithmetic mean of load data over the entire period, used to assess the overall electricity consumption level of users. Combined with the peak-valley difference parameter, it can comprehensively analyze the load fluctuation characteristics. The expression is: (1) In the formula, L t Indicates the first t Load observations for each time period; Maximum peak load C2: Used to assess the changeability and transferability of the load's electricity consumption behavior, expressed as: (2) In the formula, L max This represents the peak daily load for users. L avg This represents the average load for users during periods when they are not participating in peak shaving. Peak-to-valley ratio C3: Reflects the degree of difference between peak and valley loads, expressed as: (3) In the formula, L pp This represents the load value during peak electricity consumption periods for users. L v This refers to the load value during off-peak electricity consumption periods for users. Load volatility C4: The ratio of the standard deviation of user load to the mean during peak shaving periods, quantifying the relative intensity of load volatility, expressed as: (4) (5) In the formula, Let be the standard deviation of the load for user i; L cavg This represents the average load during the peak-shaving period for this user. m and M Do not specify the start and end time periods for peak-shaving service periods; For users i During peak service hours t The load value.

5. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 1, characterized in that, In step 3, the specific process is as follows: For soft indicators such as equipment start-up and shutdown rate (C5), electricity price sensitivity coefficient (C6), user response loss (C7), system operation and shutdown switching difficulty (C8), and system management coordination degree (C9), which are difficult to calculate directly using historical load curves, triangular fuzzy numbers (TFCs) are used. l,m,u Semantic quantization is performed, and the centroid method is used to obtain the sharpness value. The expression is as follows: (6) in, x The sharpness value after deblurring by the triangular blur number. l u k These represent the lower limit, median, and upper limit of the triangular fuzzy number, respectively; positive indicators are processed using benefit-based standardization, while negative indicators are processed using inverse standardization, resulting in a standardized evaluation matrix; The FAHP method is used to construct a fuzzy judgment matrix and calculate subjective weights. Then, the CRITIC method is used to calculate the objective weights based on the standard deviation of the indicators and the correlation between the indicators. Then, the overall weight is determined by combining weights using game theory. The overall weight and user peak-shaving potential score satisfy the following relationships: (7) (8) in, i Number the user; j The evaluation indicator number; w j For the first j The overall weight of each evaluation indicator; S i For users i Peak-shaving potential score; z ij For users i In the j Standardized values ​​under each evaluation indicator; n This represents the total number of evaluation indicators.

6. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 1, characterized in that, In step 4, the specific process is as follows: To embed the evaluation results into the virtual power plant optimization scheduling model, the user... i The peak-shaving potential score is normalized into a potential coefficient. The expression is: (9) In the formula, the lowest potential score and highest potential rating Each is taken from the set of users participating in the scheduling; Regarding power allocation, the user-reported adjustable power will be corrected to a trusted allocation limit, expressed as follows: (10) In the formula, For users i During the period t The upper limit of trusted allocation; For users i During the period t The declared adjustable power; and Assign upper bound mapping coefficients to trusted entities; For users i Potential coefficient; Regarding time-series response, the user potential coefficient is mapped to the upper limit of continuous response, expressed as: (11) In the formula, Let be the upper limit of continuous responses for user i; The upper limit of the basic continuous response; For response adjustment coefficient; In terms of economic incentives, the user potential coefficient is mapped to the valley filling subsidy price and the peak shaving subsidy price, expressed as follows: (12) (13) In the formula, and users respectively i Unit subsidy price when participating in valley filling and peak shaving services; and These are the basic subsidy prices for valley filling services and peak shaving services, respectively. γ This is the price adjustment coefficient.

7. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 1, characterized in that, In step 5, the specific process is as follows: Let the set of scheduling periods be . T The set of high-energy-consuming users participating in the scheduling is I The power grid during the period t Issue peak-shaving demand; when the demand is positive, it represents valley-filling demand, and when the demand is negative, it represents peak-shaving demand. Valley-filling demand and peak-shaving demand are defined separately, with the following expressions: (14) (15) In the formula, For the power grid during the time period t The issued peak-shaving requirements; For time period t The demand for filling the valley; For time period t Peak shaving demand; and when When positive, it indicates the demand for filling the valley. A negative value indicates peak shaving demand; With the goal of maximizing the net profit of the virtual power plant, the following objective function is established, expressed as: (16) In the formula, Net revenue from virtual power plants; , , , Let $\frac{ ... (17) (18) (19) (20) In the formula, Ω up and Ω down These are the valley filling service period set and the peak shaving service period set, respectively. and points Do not pay the unit service price for the power grid for valley filling and peak shaving services; e t up and e t down Time periods t Task shortages in the valley filling and peak shaving directions; r i up and r i down These are the unit subsidy prices for user i when participating in valley filling and peak shaving services, respectively. x i,t up and x i,t down Let represent the power increase and power decrease for user i during time period t, respectively; I represents the set of high-energy-consuming users participating in the scheduling. p t ch and p t dis These represent the charging power and discharging power of the energy storage during time period t, respectively. c es The charging and discharging cost of energy storage units; T For the set of scheduling periods; Cmiss The cost of penalties for shortfall; The penalty coefficient per unit of shortfall; and Time periods t There is a shortage of tasks in the valley filling and peak reduction directions.

8. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 1, characterized in that, The virtual power plant optimization scheduling model includes at least the following constraints: (1) User load reconfiguration constraints: The load after dispatch is determined by the baseline load, the upward adjustment power, and the downward adjustment power, and its expression is: (21) in, For users i During the period t The actual load after scheduling; For users i During the period t The baseline load; For users i During the period t Increased power; For users i During the period t The power reduction; (2) Trusted Allocation Upper Limit Constraint: The power increase and decrease of a user shall not exceed its trusted allocation upper limit, the expression of which is: (22) in, For users i During the period t The upper limit of trusted allocation; and users respectively i During the period t The upward and downward adjustment of power; (3) Continuous response upper limit constraint: The change in net regulation power between adjacent time periods shall not exceed the user's continuous response upper limit, the expression of which is: (23) in, For users i The upper limit of continuous response; and Representing users respectively i During the period t Net regulating power; and Representing users respectively i Net regulation power in the previous period; (4) Daily power balance constraint: The deviation of the total daily power consumption before and after peak shaving shall not exceed the allowable range, and its expression is: (24) in, For users i Allowable daily power consumption deviation within the scheduling cycle; T For the set of scheduling periods; and users respectively i During the period t The upward and downward adjustment of power; (5) Energy storage power constraint: Both the charging power and discharging power of energy storage must meet the rated power boundary, the expression of which is: (25) (26) in, and These represent the charging power and discharging power of the energy storage during time period t, respectively. and These are the maximum charging power and maximum discharging power of the energy storage, respectively. (6) Energy storage state of charge constraint: The energy storage state of charge changes dynamically according to the charging and discharging power, and always meets the upper and lower limit constraints. Its expression is: (27) (28) in, and Energy storage time period t The state of charge in the previous period; and These refer to energy storage charging efficiency and discharging efficiency, respectively. and These are the lower and upper limits of the energy storage state of charge, respectively; This refers to the rated capacity of the energy storage. (7) Peak shaving task tracking constraints: During the valley filling period, the task is jointly undertaken by industrial users increasing their load and energy storage charging; during the peak shaving period, the task is jointly undertaken by industrial users decreasing their load and energy storage discharging. The expression is as follows: (29) (30) in, and Time periods t Valley filling and peak shaving demand; and These are the task vacancies in the corresponding directions; and These are the energy storage charging power and the discharging power, respectively. and These options allow users to increase or decrease power, respectively. (8) System power balance constraint: Purchased electricity is used as a system balancing item to compensate for the net power difference after internal resource adjustment. Its expression is: (31) in, The purchased power during time period t; For time period t Other loads; For time period t The output of new energy sources; For time period t Small conventional power supply output; For users i During the period t The actual load after scheduling; and These are the energy storage charging power and the discharging power, respectively. (9) Individual rationality constraint: The net energy cost of users participating in peak shaving is not higher than that of the baseline scenario, and its expression is: (32) in, For users i Net energy cost in scheduling scenarios; For users i Net energy cost in the baseline scenario.

9. The virtual power plant load optimization scheduling method considering the peak-shaving potential of high-energy-consuming users according to claim 1, characterized in that, In step 6, the specific process is as follows: A linear programming solver is used to solve the problem. When the mutual exclusion state of energy storage charging and discharging or the user start-up and shutdown state are introduced, a mixed integer linear programming solver is used. After the solution is completed, the output is the time-sharing adjustment power of each user, the energy storage charging and discharging power and SOC trajectory, the service window task deficit, the virtual power plant service revenue, the user subsidy expenditure, the energy storage operating cost, the virtual power plant net revenue and the net energy cost of each user.

Images