Two-stage game operation optimization method of virtual power plant considering real-time demand response

Abstract

The application discloses a kind of virtual power plant two-stage game operation optimization method considering real-time demand response, comprising the following steps: S1, establishing two-stage game optimization model, the two-stage game optimization model includes day-ahead stage energy supply side scheduling optimization model and day-in stage energy deviation coordination optimization model;S2, initialize two-stage game optimization model;S3, in two-stage game optimization model, day-ahead stage energy supply side scheduling optimization model obtains the energy scheduling scheme of each DERs in day-ahead, the power purchase scheme of VPP control center by solving the objective function of day-ahead DERs and virtual power plant control center;The energy deviation coordination optimization model of day-in stage uses iterative algorithm, obtains the real-time power consumption strategy of user minimizing energy deviation.This method obtains the energy scheduling scheme of DERs in day-ahead stage VPP energy supply side and the real-time power consumption strategy of user minimizing energy deviation in day-in stage through two-stage optimization model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual power plant operation, and particularly refers to a two-stage game operation optimization method for virtual power plant considering real-time demand response. BACKGROUND

[0002] With the rapid growth of urban population and the rapid development of urban economy, energy consumption has increased significantly, which brings great pressure to the operation of urban power system. On the one hand, the imbalance between power supply and demand will lead to instability or even collapse of urban power system. On the other hand, peak load may lead to high electricity prices, which will increase the energy cost of end users. The traditional solution is to build centralized power plants to alleviate urban power shortage or imbalance between supply and demand. This method is too expensive and will cause a large amount of carbon emissions. In order to reduce pollution of traditional thermal power plants and at the same time alleviate urban power shortage or imbalance between supply and demand, the proportion of distributed energy resources (DERs) generation is increasing. However, DERs are usually distributed in small capacity and have fluctuating output. In contrast, urban virtual power plant (VPP) can aggregate various DERs, including wind energy, photovoltaic and energy storage systems, through advanced information and communication technology. Urban VPP makes full use of these aggregated energy sources, and through optimal scheduling of traditional energy and renewable energy, it realizes complementary energy supply on the supply side and also can improve the utilization rate of renewable energy. In addition, through the coordination of DERs, urban VPP can participate in the operation and dispatch of power system as a whole. Therefore, it is an important problem to realize the economic operation of urban VPP and to realize the optimal scheduling of DERs in urban VPP.

[0003] Participation of demand response (DR) is a mainstream initiative to alleviate the power supply pressure in VPP, and can also reduce the cost of users using energy. Different users usually have different needs and responses when participating in DR. Therefore, it is particularly important to develop more flexible DR to provide different incentives for different users. In addition, both DERs and users have individual rationality and compete with each other in conflict. In the process of VPP aggregating each distributed resource to participate in DR optimization, how to use game interaction to make fair and just revenue sharing decisions is beneficial to encourage each distributed resource to participate in DR, and is of great significance to VPP participating in demand response of power grid company. The uncertainty problem of VPP user energy use in a day further aggravates the energy deviation problem in the energy market, which brings unavoidable energy deviation to the user's day-ahead and real-time trading process, and directly affects the stability and economy of VPP operation. In view of the energy deviation problem in the energy market, the mainstream solution at present includes improving the accuracy of energy use prediction, introducing intermediaries and market adjustment mechanism, etc., but the current methods for handling energy deviation have deficiencies in balancing the interests of the supply side and the demand side in the market, and cannot fully coordinate the interests of all parties. The flexibility of user DR resource regulation can effectively reduce the energy deviation from the load side, but considering the regulation constraints of DR, most of the current researches realize DR regulation in the day-ahead stage, and the real-time regulation strategy of day-ahead DR needs to be further discussed. Therefore, the analysis of the VPP operation optimization problem considering real-time demand response will provide certain guiding significance for the distributed autonomy and collaborative optimization of VPP. SUMMARY

[0004] The present application is directed to the deficiencies of the prior art, and proposes a virtual power plant two-stage game operation optimization method considering real-time demand response. The method obtains the energy scheduling scheme of DERs on the energy supply side of VPP in the day-ahead stage and the real-time power consumption strategy of users for minimizing energy deviation in the day-ahead stage through a two-stage optimization model.

[0005] In order to solve the above technical problems, the technical scheme of the present application is as follows:

[0006] A VPP two-stage game operation optimization method considering real-time demand response, comprising the following steps:

[0007] S1, a two-stage game optimization model of VPP considering real-time demand response is established, and the two-stage game optimization model includes a day-ahead stage energy supply side scheduling optimization model and an intra-day stage energy deviation coordination optimization model.

[0008] S2, the two-stage game optimization model of VPP is initialized, and the parameters of the two-stage game optimization model include the operation cost coefficient β of the energy storage system (ESS), the charge and discharge power coefficient ρ of the ESS ch , ρ dis ; the optimization time step Δt; the capacity of the ESS ESS charging and discharging maximum power GT cost coefficient k GT ; average electricity purchase price λ provided by the market to DERs mk ; user electricity load reduction limit coefficient ε cut ; user electricity load reduction discomfort coefficient μ cut,i ; user electricity load transfer discomfort coefficient μ tr,i ; transferable load operation rated power P rated ; unit penalty cost of energy deviation

[0009] S3, in the two-stage game operation optimization model of VPP, the day-ahead energy supply side scheduling optimization model adopts a distributed method combining adaptive differential evolution algorithm and Gurobi solver to solve the objective function of the day-ahead DERs and VPP control center, and obtain the energy scheduling scheme of each DER and the electricity purchase scheme of the VPP control center. The energy deviation coordination optimization model in the intraday stage adopts an iterative algorithm to obtain the user's real-time electricity strategy that minimizes the energy deviation.

[0010] As preferred, the day-ahead energy supply side scheduling optimization model in S1 is a double-layer optimization model based on stackelberg master-slave game, wherein the bottom layer follower is each DER, photovoltaic power generation, wind power generation and gas turbine are selected, and all devices are equipped with energy storage system; the upper layer leader is the VPP control center.

[0011] The energy storage system equipped by each DER is modeled as follows:

[0012]

[0013] wherein, is the operating cost of the energy storage device; is the charging and discharging power of the energy storage device at time t; is the amount of electricity stored by ES at time t.

[0014] To avoid full charging and discharging of the energy storage system and reduce battery life, the battery capacity should be between 20% and 80% of the rated capacity, and cannot be charged and discharged at the same time, with the following specific constraints:

[0015]

[0016] wherein, respectively represent the maximum power of charging and discharging; is a 0-1 variable, respectively representing the charging state and discharging state of the ESS at time t.

[0017] As preferred, in the energy supply side scheduling optimization model of the day-ahead stage, the photovoltaic generator set takes its own benefit F PV Maximization is the optimization objective, and the objective function is:

[0018]

[0019] Wherein, λ t is the internal selling price of VPP; is the total power traded between PV and VPP control center at t time; is the charging and discharging power of the energy storage equipped for PV; is the operation cost of the ES equipped for PV; is the predicted power generation of PV.

[0020] As preferred, in the energy supply side scheduling optimization model of the day-ahead stage, the wind turbine generator set takes its own benefit F WT Maximization is the optimization objective, and the objective function is:

[0021]

[0022] Wherein, is the total power traded between WT and VPP control center at t time; is the charging and discharging power of the energy storage equipped for WT; is the operation cost of the ES equipped for WT; is the predicted power generation of WT.

[0023] As preferred, in the energy supply side scheduling optimization model of the day-ahead stage, the gas turbine takes its own benefit F GT Maximization is the optimization objective, and the objective function is:

[0024]

[0025] In the formula, is the total power traded between GT and VPP control center at t time; is the output electric power of GT at t time in the day-ahead energy production scheduling; is the charging and discharging power of the energy storage equipped for GT; is the operation cost of the ES equipped for GT. is the operation cost of GT.

[0026] As preferred, the upper leader VPP control center takes the VPP power purchase cost C VPP Minimization is the optimization objective, and the objective function is:

[0027]

[0028] Wherein, CVPP the electricity purchase cost of the VPP control center; the contracted electricity quantity of the VPP control center and the grid company; the contracted electricity price; the user-side load predicted power at time t; the total output power of the DER inside the VPP at time t.

[0029] To avoid the VPP control center lowering the electricity price to the minimum for maximizing its own benefits, the energy supply side dispatch optimization model in the day-ahead stage sets relevant electricity price constraints to help the electricity price adjust within a certain range, while the average value of the electricity price throughout the day is not lower than the market electricity price. The specific constraints are as follows:

[0030]

[0031] As an option, the energy deviation coordination optimization model in S1 is a cooperative alliance composed of multiple users, and each user aims to minimize the electricity cost with the objective function as follows:

[0032]

[0033] wherein, is the total electrical load of user i at time t in the day; is the discounted electricity price obtained by the user participating in the cooperative alliance; is the discomfort cost of user i caused by the reducible load at time t; is the discomfort cost of user i caused by the transferable load at time t; is the incentive income of user i participating in the cooperative alliance at time t; are the basic electrical load, the reducible electrical load, and the transferable electrical load of user i at time t in the day, respectively; is the external market electricity price at time t.

[0034] As an option, the user reducible load regulation range constraint is set as follows:

[0035]

[0036] As an option, to ensure the completion of the daily transferable load task, the relevant capacity constraints are set as follows:

[0037]

[0038] wherein, E i is the total amount of transferable load that user i needs to complete within 24 hours; Δt is the decision step in the day, which is 30 minutes in this study. When making decisions at time t in the day, user i needs to make the following judgments:

[0039]

[0040] If (I) is true, it means that the user i has completed the daily transferable load task, and the transferable load at time t and the subsequent time of the day is zero. If (II) is true, it means that the transferable load at time t and the subsequent time of the day needs to be run at full power to complete the daily transferable load task. In other cases, the user i can freely regulate the start and stop of the transferable load as needed.

[0041] As preferred, the coalition adopts the Shapley value method to allocate the incentive subsidies to the users. The allocation rule is as follows:

[0042]

[0043] wherein N is a set of participants in the coalition, representing {user 1, user 2, …, user i}; is the benefit obtained by the coalition member e; φ(S) is the weight of the coalition member sharing the benefits; S\{i} is the set after excluding member i from the set S; e(S) is the characteristic function of the coalition; v(S) is the total benefit of the coalition; x i is the income of member i before participating in the cooperative game; S is a set of different coalition combinations.

[0044] As preferred, the intra-day stage cooperative coalition takes the electricity cost as the optimization objective, and the objective function is:

[0045]

[0046] wherein, is the penalty cost of the power deviation of the intra-day load.

[0047] As preferred, the step S3 comprises the following sub-steps:

[0048] S3-1, initialize the total load power prediction value of the user side (i.e., the sum of all user loads of the user side on the next day, and the initial value is set according to the actual prediction situation) and the power generation prediction power of each DER in the day-ahead stage.

[0049] S3-2, the model is solved by using the distributed algorithm of ADE nested Gurobi solver in the day-ahead stage. The VPP control center transmits the initialized internal power purchase price of the VPP to each DER in the lower layer, and the DER in the lower layer respectively optimizes the energy selling scheme according to the prediction of its own power generation power by using the Gurobi solver to obtain the power output result of the lower layer and transmits it to the VPP control center in the upper layer, and the VPP control center calculates its benefit value C VPP, and update the electricity price, which is then passed to the lower layer for optimization. Until the ADE algorithm converges, the optimal VPP energy acquisition plan and DERs energy scheduling plan are obtained, and the energy acquisition plan is passed to the intraday stage for optimization.

[0050] In the intraday phase, real-time demand response and cooperative game play are considered moment by moment. At time t, the VPP control center determines whether the load is within the ±2% permissible energy deviation range based on the day-ahead power purchase plan and the current real-time load. If within the permissible energy deviation range, no energy deviation penalty is imposed at the current moment, and no action is taken, proceeding to the next moment. If the permissible energy deviation range is exceeded, an iterative algorithm is used to solve the energy deviation collaborative optimization model based on the cooperative game to determine the user's real-time electricity usage strategy that minimizes the energy deviation.

[0051] Preferably, the step S3-2 includes the following sub-steps:

[0052] S3-2-1. Initialize the simulation parameters of the energy supply side scheduling optimization model based on the Stackelberg master-slave game in the day-ahead phase in S1. The simulation parameters include the operating cost coefficient β of the energy storage system; the charging and discharging power coefficient ρ of the ESS. ch , ρ dis ;Optimize time step Δt; ESS capacity Maximum power of ESS charging and discharging GT cost coefficient k GT ; The average electricity purchase price provided by the market to DERs λ mk ;The maximum number of iterations K of the ADE algorithm max ; Population size n size ; The dimension n of each individual in the population dim ; Search interval for target solution The scaling factor F of population variation and the crossover recombination probability theory CR.

[0053] S3-2-2. Initialize population a, i.e. randomly generate n populations within the VPP internal electricity price control constraints. size The energy purchase price is used as the initial population, and the number of iterations is initialized to K = 0.

[0054] S3-2-3: The VPP internal energy purchase price in population a is transmitted to each DER in the lower layer. Each DER uses the Gurobi solver according to the electricity price to optimize its power output. Upload the optimization results to the upper-level leader VPP control center.

[0055] S3-2-4. The VPP control center calculates the benefit value of each individual in population a based on the optimization results of the lower layer, and takes the maximum benefit value E1 of this round.

[0056] S3-2-5, variation and crossover of the population a to obtain a new population b.

[0057] S3-2-6, the VPP internal energy procurement price in the population b is transmitted to each DER in the lower layer, the DERs again call the Gurobi solver to optimize the objective function, and the optimization result is transmitted to the upper layer, the VPP control center calculates the benefit value of each individual in the population b according to the returned lower layer optimization result, and takes out the maximum benefit value E2 of this round.

[0058] S3-2-7, selection operation, if E2>E1, then a=b, E1=E2, if E2<E1, keep unchanged.

[0059] S3-2-8, judgment operation, if the maximum iteration number (K=K max ) is reached, the optimization result is output, otherwise jump to S3-2-5.

[0060] As preferred, the step S3-3 comprises the following sub-steps:

[0061] S3-3-1, initialize the simulation parameters of the intra-day stage energy deviation coordination optimization model in S1, the simulation parameters include the limiting coefficient ε cut,i of the load reduction of each user, the inadaptation coefficient μ cut,i of the load reduction and load transfer of each user, the rated power P tr,i of the transferable load, the unit penalty cost of the energy deviation rated

[0062] S3-3-2, at time t, compare the real-time load at the current time and the purchase electricity plan at the same time in the day-ahead purchase electricity scheme, judge whether the load deviation is within the allowable power deviation range of ±2%, that is If it is within the allowable power deviation range, there is no power deviation penalty at the current time, no operation is performed, and it is transferred to S3-3-10; otherwise, it is transferred to S3-3-3.

[0063] S3-3-3, at time t, the user-side load deviation exceeds the allowable range, the VPP control center issues a load power adjustment call, each user adjusts the flexible load demand to the maximum extent within the respective load adjustment range, and obtains the electricity consumption strategy of each user

[0064] S3-3-4, calculate the incentive subsidies of each user according to the Shapley value method and calculate the total energy cost of each user

[0065] ​S3-3-5, judging whether the power deviation exceeds the allowable deviation range, if exceeding the range, it indicates that the maximum load response at t time cannot completely eliminate the energy deviation, the game stops, and the power consumption strategy of each user is output Go to S3-3-10; otherwise go to S3-3-6.

[0066] S3-3-6, the load power (the change amount of the load that can be cut or transferred) of each user is halved in the respective load adjustment range, and the power consumption strategy of each user is obtained

[0067] S3-3-7, judging whether the current round power deviation exceeds the allowable deviation range, if exceeding the range, the power consumption strategy of each user is output Go to S3-3-10; otherwise go to S3-3-8.

[0068] S3-3-8, according to the power consumption strategy the incentive subsidy of each user is calculated and the total energy cost of each user is calculated the and of each user are compared if the of the user i in the strategy B is updated the power consumption strategy of the user i in the strategy A is updated; otherwise the power consumption strategy of the user i is unchanged.

[0069] S3-3-9, judging whether all users meet if meeting, the DR strategy of all users is determined, the game ends, and the power consumption strategy of the user is output Go to S3-3-10; otherwise go to S3-3-6.

[0070] S3-3-10, the power deviation cooperative optimization at t time ends, and waiting for entering t+1 time.

[0071] The application has the following characteristics and beneficial effects:

[0072] By adopting the technical scheme, the application establishes a day-ahead-day-in two-stage game optimization model to solve the decision and optimization problems among multiple subjects in the VPP, and realizes the day-ahead economic optimization scheduling and the day-in energy deviation collaborative optimization of the VPP. In the day-ahead stage, the benefit distribution of the market is effectively balanced through benign competition, and on the basis of ensuring the balanced DERs income, the power purchase cost of the VPP control center is effectively reduced. In the day-in stage, while ensuring the economy and satisfaction of the user power consumption, the energy deviation of the load side in the day-in is effectively reduced, and the waste of resources is prevented. At the same time, the user can be better motivated to use electricity economically and reasonably, the reliability of power supply is ensured, the distributed autonomy and collaborative optimization of the VPP are realized, the dependence of the VPP on the power grid is reduced, and the power supply pressure of the power grid is relieved. BRIEF DESCRIPTION OF DRAWINGS

[0073] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0074] Figure 1 The figure is a system architecture diagram of the VPP two-stage operation optimization method of the embodiment of the present application.

[0075] Figure 2 The figure is the optimization result of the gas turbine in the day-ahead stage in the embodiment of the present application.

[0076] Figure 3 The figure is the optimization result of the photovoltaic generator set in the day-ahead stage in the embodiment of the present application.

[0077] Figure 4 The figure is the optimization result of the wind turbine generator set in the day-ahead stage in the embodiment of the present application.

[0078] Figure 5 The figure is the optimization result of the internal power purchase price of the VPP in the day-ahead stage in the embodiment of the present application.

[0079] Figure 6 The figure is the load comparison before and after the optimization in the day-in stage.

[0080] Figure 7 The figure is the power grid power purchase power comparison before and after the optimization in the day-in stage.

[0081] Figure 8 The figure is the VPP power deviation penalty in the day-in stage in the embodiment of the present application. DETAILED DESCRIPTION

[0082] It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0083] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings.

[0084] On the contrary, the present invention covers any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention as defined by the claims. Furthermore, to facilitate a better understanding of the present invention, certain specific details are described in detail below in the detailed description of the present invention. Those skilled in the art will be able to fully understand the present invention without these details.

[0085] The present invention provides a two-stage game operation optimization method for a virtual power plant considering real-time demand response, which involves a VPP operation optimization method. A two-stage game optimization model is used to generate a VPP day-ahead economic optimization scheduling plan and an intraday electricity deviation collaborative optimization strategy. The two-stage optimization models are an energy supply-side scheduling optimization model for the day-ahead stage and an energy deviation coordinated optimization model for the intraday stage. The two stages solve the optimal scheduling plan for day-ahead DERs and the real-time electricity consumption strategy for intraday users, respectively, as follows: Figure 1 As shown. The purpose of the present invention is to propose a two-stage game operation optimization method for VPPs that takes into account real-time demand response, so as to enable DERs on the VPP energy supply side to obtain more production capacity benefits, effectively reduce the energy deviation on the load side during the day, operate urban VPPs at the lowest cost, realize the distributed autonomy and collaborative optimization of VPPs, reduce the dependence of VPPs on the power grid, and alleviate the power supply pressure of the power grid. To illustrate the effect of the present invention, the present invention is described in detail below using a certain VPP as the implementation object of the present invention, including the following steps:

[0086] S1. Establish a VPP two-stage game optimization model that considers real-time demand response. The two-stage game optimization model includes an energy supply-side scheduling optimization model in the day-ahead stage and an energy deviation coordination optimization model in the intraday stage.

[0087] The two-stage optimization model is Figure 1 shown.

[0088] The energy supply-side dispatch optimization model for the day-ahead phase is a two-layer optimization model based on the Stackelberg master-slave game. The bottom-layer followers are the DERs, which select photovoltaic power generation, wind power generation, and gas turbines, and consider that all equipment is equipped with energy storage systems; the top-layer leader is the VPP control center.

[0089] The unified modeling of the energy storage system equipped with each DER is as follows:

[0090]

[0091] in, The operating cost of the energy storage device; The charging and discharging power of the energy storage device at time t; The amount of electricity stored by the ES at time t.

[0092] To avoid full charging and discharging of the energy storage system and reduce the battery life, the battery capacity should be between 20% and 80% of the rated capacity, and cannot be charged and discharged at the same time, with the following specific constraints:

[0093]

[0094] Wherein, respectively represent the maximum power of charging and discharging; are 0-1 variables, respectively representing the charging state and discharging state of the ES at time t.

[0095] The photovoltaic generator set maximizes its own benefit F PV as the optimization objective, and the objective function is:

[0096]

[0097] Wherein, λ t is the internal electricity price of the VPP; is the total power traded between the PV and the VPP control center at time t; is the charging and discharging power of the energy storage equipped for the photovoltaic power station; is the operating cost of the ES equipped for the PV; is the predicted power generation of the PV.

[0098] The wind turbine maximizes its own benefit F WT as the optimization objective, and the objective function is:

[0099]

[0100] Wherein, is the total power traded between the WT and the VPP control center at time t; is the charging and discharging power of the energy storage equipped for the WT; is the operating cost of the ES equipped for the WT; is the predicted power generation of the WT.

[0101] The gas turbine maximizes its own benefit F GT as the optimization objective, and the objective function is:

[0102]

[0103]

[0104] In the formula, is the total power traded by GT with VPP control center at time t; is the output power of GT at time t in day-ahead power production dispatching; is the charge-discharge power of ES equipped for GT; is the operation cost of ES equipped for GT. is the operation cost of GT;

[0105] VPP control center in day-ahead stage to purchase power at cost C VPP The optimization objective is to minimize the objective function:

[0106]

[0107] wherein C VPP is the power purchase cost of VPP control center; is the contracted power of VPP control center with grid company; is the contracted power price; is the user-side load forecast power at time t; is the total output power of DER inside VPP at time t.

[0108] To avoid VPP control center to minimize the price for its own benefit maximization, the model sets relevant price constraints to help the price adjust within a certain range, while the average value of the whole day price is not lower than the market price. The specific constraints are as follows:

[0109]

[0110] Ⅱ) The intra-day stage energy deviation coordination optimization model is a cooperative alliance composed of multiple users, and each user aims to minimize the electricity cost The optimization objective is to minimize the objective function:

[0111]

[0112] wherein, is the total electrical load of user i at time t in intra-day; is the discounted price obtained by user participating in the cooperative alliance; is the discomfort cost of user i at time t caused by load reduction; is the discomfort cost of user i at time t caused by load shifting; is the incentive income obtained by user i participating in the cooperative alliance at time t; are the basic electrical load, the reducible electrical load and the shiftable electrical load of user i at time t in intra-day, respectively; is the external market price at time t.

[0113] The user can reduce the load control range constraint settings as follows:

[0114]

[0115] To ensure the completion of daily transferable load tasks, the relevant capacity constraints are set as follows:

[0116]

[0117] Among them, E i is the total amount of transferable load that user i needs to complete within 24 hours; Δt is the decision step within the day, which is 30 minutes in this study. When making a decision at time t within the day, user i needs to make the following judgments:

[0118]

[0119] If the result of (I) is true, it means that user i has completed the transferable load task for the day, and the transferable load at time t and the following time of the day is zero. If the result of (II) is true, it means that the transferable load at time t and the following time of the day must run at full power to complete the transferable load task for the day. In other cases, user i can freely control the start and stop of the transferable load as needed.

[0120] The Alliance uses the Shapley value method to allocate incentive allowances to users. The allocation rules are as follows:

[0121]

[0122] Where N is the set of participants in the alliance, represented as {user 1, user 2, ..., user i}; is the benefit obtained by alliance member e; φ(S) is the weight of the benefit shared by alliance members; S\{i} is the set after member i is excluded from set S; e(S) is the characteristic function of the alliance; v(S) is the total benefit of the alliance; x i is the income of member i before participating in the cooperative game; S is the set of different alliance combinations.

[0123] The intraday phase of the cooperative alliance is based on electricity costs Minimum is the optimization goal, and its objective function is:

[0124]

[0125] in; The penalty cost caused by the power deviation of the daily load.

[0126] S2. Initialize the VPP two-stage game operation optimization model. The optimization model parameters include the energy storage system (ESS) operation cost coefficient β; the ESS charge and discharge power coefficient ρ ch , ρdis ;Optimize time step Δt; ESS capacity Maximum power of ESS charging and discharging GT cost coefficient k GT ; The average electricity purchase price provided by the market to DERs λ mk ; Limitation coefficient of user electricity load reduction ε cut ; User load reduction discomfort coefficient μ cut,i ; User load transfer discomfort coefficient μ tr,i ; Rated power P for transferable load operation rated ; Unit penalty cost for energy deviation Unit penalty cost for energy deviation

[0127] S3. In the VPP two-stage game operation optimization model, the energy supply-side scheduling optimization model in the day-ahead phase uses a distributed method that combines an adaptive differential evolution algorithm with a Gurobi solver to solve the objective functions of the day-ahead DERs and the VPP control center, obtaining the day-ahead energy scheduling plan for each DER and the power purchase plan for the VPP control center. The energy deviation coordination optimization model in the intraday phase uses an iterative algorithm to obtain the user's real-time power consumption strategy that minimizes the energy deviation. This specifically includes the following sub-steps:

[0128] S3-1. Initialize the forecast value of the total user-side load power on the previous day (i.e., the sum of all user loads on the user-side for the next day, with the initial value set according to the actual forecast situation) and the forecast power generation of each DER on the previous day.

[0129] S3-2, in the day-ahead stage, the model is solved using a distributed algorithm that nests the ADE Gurobi solver. The VPP control center passes the initialized VPP internal electricity purchase price to each DER in the lower layer. The DERs in the lower layer use the Gurobi solver to optimize their respective energy sales plans based on their own power generation forecasts, obtain the power output results of the lower layer, and pass them to the upper-layer VPP control center. The VPP control center uses the ADE algorithm to calculate its benefit value, update the electricity price, and pass it to the lower layer for optimization. Until the ADE algorithm converges, the optimal VPP energy acquisition plan and DERs energy scheduling plan are obtained, and the energy acquisition plan is passed to the intraday stage for optimization. The specific sub-steps are as follows:

[0130] S3-2-1. Initialize the simulation parameters of the energy supply side scheduling optimization model based on the Stackelberg master-slave game in the day-ahead phase in S1. The simulation parameters include the operating cost coefficient β of the energy storage system; the charging and discharging power coefficient ρ of the ESS. ch , ρ dis ;Optimize time step Δt; ESS capacity Maximum power of ESS charging and discharging Cost coefficient k of GT GT ; average electricity purchase price λ provided by the market to DERs mk ; maximum iteration number K of ADE algorithm max ; population number n size ; dimension n of each individual in the population dim ; search interval of target solution Scaling factor F of population mutation and crossover recombination probability CR.

[0131] S3-2-2, initialize population a, that is, randomly generate n energy purchase prices within the VPP internal electricity price regulation constraints as the initial population, and initialize the iteration number K = 0. size

[0132] S3-2-3, transmit the VPP internal energy purchase price in population a to each DER in the lower layer, and each DER calls the Gurobi solver according to the electricity price to respectively optimize its own power output The optimization result is uploaded to the upper leader VPP control center.

[0133] S3-2-4, the VPP control center calculates the benefit value of each individual in population a according to the lower layer optimization result, and takes out the maximum benefit value E1 of this round.

[0134] S3-2-5, mutate and cross population a to get new population b.

[0135] S3-2-6, transmit the VPP internal energy purchase price in population b to each DER in the lower layer, and the DERs again call the Gurobi solver to optimize its objective function, and transmit the optimization result to the upper layer. The VPP control center calculates the benefit value of each individual in population b according to the returned lower layer optimization result, and takes out the maximum benefit value E2 of this round.

[0136] S3-2-7, selection operation, if E2 > E1, then a = b, E1 = E2, if E2 < E1, keep unchanged.

[0137] S3-2-8, judgment operation, if the maximum iteration number (K = K max ) is reached, the optimization result is output, otherwise jump to S3-2-5.

[0138] ​S3-3, consider real-time demand response and cooperative game at each time in the day, VPP control center judges whether the load is within the allowable power deviation range of ±2% according to the day-ahead power purchase scheme and the real-time load at the t time. If it is within the allowable power deviation range, there is no power deviation penalty at the current time, no operation is performed, and the next time is entered. If it exceeds the allowable power deviation range of ±2%, an iterative algorithm is used to solve the energy deviation cooperative optimization model based on cooperative game to obtain the user real-time power consumption strategy that minimizes the energy deviation. Specifically, the following sub-steps are as follows:

[0139] S3-3-1, initialize the simulation parameters of the energy deviation coordination optimization model in the day in S1, the simulation parameters include the limiting coefficient ε of the reduction of the electric load of each user cut,i ; the inadaptation coefficient μ of the reduction of the electric load and the transfer of the electric load of each user cut,i , μ tr,i ; the rated power P of the transferable load rated ; the unit penalty cost of the energy deviation unit penalty cost

[0140] S3-3-2, at the t time, compare the real-time load at the current time with the power purchase plan at the same time in the day-ahead power purchase scheme to judge whether the load deviation is within the allowable power deviation range of ±2%, i.e. If it is within the allowable power deviation range, there is no power deviation penalty at the current time, no operation is performed, and S3-3-10 is turned to; otherwise, S3-3-3 is turned to.

[0141] S3-3-3, at the t time, the user-side load deviation exceeds the allowable range, the VPP control center issues a load power adjustment call, each user adjusts the flexible load demand to the maximum extent within the respective load adjustment range, and obtains the user power consumption strategy

[0142] S3-3-4, calculate the incentive allowance of each user according to the Shapley value method and calculate the total energy cost of each user

[0143] S3-3-5, judge whether the power deviation exceeds the allowable deviation range, if it exceeds the range, it means that the maximum load response at the t time cannot completely eliminate the energy deviation, the game stops, and the user power consumption strategy is output S3-3-10 is turned to; otherwise, S3-3-6 is turned to.

[0144] S3-3-6, halve the load power response (change amount of the reducible load and the transferable load) of each user within the respective load adjustment range to obtain the user power consumption strategy

[0145] S3-3-7. Determine whether the current round power deviation exceeds the allowable deviation range. If it exceeds the range, output the user's power usage strategy Go to S3-3-10; otherwise go to S3-3-8.

[0146] S3-3-8. According to the electricity usage strategy Calculate incentive allowances for each user And calculate the total energy cost of each user Compare each user's and like Use renew Using policy B, user i Update the electricity consumption strategy of user i in strategy A; otherwise User i's electricity usage strategy remains unchanged.

[0147] S3-3-9. Determine whether all users meet If satisfied, all users determine their own DR strategies, the game ends, and the user power consumption strategy is output. Go to S3-3-10; otherwise go to S3-3-6.

[0148] S3-3-10, the collaborative optimization of power deviation at time t is completed, and the process is waiting to enter time t+1.

[0149] It should be noted that during the optimization process, the optimization results of each DER in the day-ahead phase are as follows: Figure 2 、 Figure 3 、 Figure 4 As shown; the optimization results of the power price on the VPP internal energy supply side are as follows Figure 5 As shown; the operating costs and benefits of each DER in the day-ahead game bidding and the costs and benefits of DER in the conventional power market are shown in Table 1 and Table 2, and the comparison of the energy purchase costs of the VPP control center in the conventional market and the VPP internal market is shown in Table 3. The above results can reflect that DERs with different characteristics in the day-ahead model can be effectively optimized and regulated under the guidance of the power purchase price of the VPP control center, and the best optimization strategy can be selected to participate in market transactions, ensuring that the load demand is met while effectively improving the economic benefits of DERs and VPP control centers. The comparison of intraday load and power grid purchase volume is shown in Figure 6 、 Figure 7 As shown in the figure, the penalty situation caused by VPP power default before and after load demand response optimization is compared. Figure 8The 24h electricity cost and benefit data analysis of the user before and after the load demand response optimization is shown in Table 4. From the above results, the energy deviation collaborative optimization model considering real-time demand response and cooperative game at each time in the intra-day stage can effectively reduce the energy deviation of the load side in the intra-day, and reduce the energy deviation penalty by 83.96%, while ensuring the electricity economy and satisfaction of the user.

[0150] Table 1 DER cost and benefit analysis in VPP internal market

[0151]

[0152] Table 2 DER cost and benefit in conventional electricity market

[0153]

[0154] Table 3 Comparison of energy purchasing cost of VPP control center

[0155]

[0156] Table 4 Comparison of 24h electricity cost and benefit of user

[0157]

[0158]

[0159] The embodiments of the present application are described in detail above with reference to the drawings, but the present application is not limited to the described embodiments. For those skilled in the art, various changes, modifications, replacements and variations of the embodiments including components can be made without departing from the principles and spirits of the present application, and still fall within the protection scope of the present application.

Claims

1. A two-stage game operation optimization method for a virtual power plant considering real-time demand response, characterized in that: The steps include: S1. Establish a two-stage game optimization model, which includes a day-ahead energy supply-side scheduling optimization model and an intraday energy deviation coordination optimization model; the day-ahead energy supply-side scheduling optimization model is a two-layer optimization model based on the Stackelberg master-slave game; the intraday energy deviation coordination optimization model is an energy deviation coordination optimization model based on the cooperative game; In the two-layer optimization model based on the Stackelberg master-slave game, the bottom followers are the DERs, which select photovoltaic power generation, wind power generation, and gas turbines, all of which are equipped with energy storage systems; the top leader is the VPP control center; In the intra-day energy deviation coordination optimization model, each user uses electricity cost Minimization is the optimization goal, and its objective function is: in, is the total electricity load of user i at time t within the day; Discounted electricity prices for users participating in cooperative alliances; is the discomfort cost caused by load reduction for user i at time t; is the discomfort cost caused by the transferable load of user i at time t; is the incentive benefit obtained by user i from participating in the cooperative alliance at time t; are the basic load, curtailable load, and transferable load of user i at time t within the day; is the external market electricity price at time t, ε cut,i is the limiting coefficient of user power load reduction, μ tr,i The discomfort coefficient of user's electric load transfer; S2. Initialize a two-stage game optimization model. The parameters of the two-stage game optimization model include the operating cost coefficient β of the ESS; the charging and discharging power coefficient ρ of the ESS ch , ρ dis ;Optimize time step Δt; ESS capacity Maximum power of ESS charging and discharging GT cost coefficient k GT DERs’ average electricity purchase price λ mk ; Limitation coefficient of user electricity load reduction ε cut,i ; User load reduction discomfort coefficient μ cut,i ; User load transfer discomfort coefficient μ tr,i ; Rated power P for transferable load operation rated ; Unit penalty cost for energy deviation S3. In the two-stage game optimization model, the day-ahead energy supply-side scheduling optimization model uses a distributed approach combining an adaptive differential evolution algorithm with a Gurobi solver to solve the objective functions of the day-ahead DERs and the virtual power plant control center, thereby obtaining the day-ahead energy scheduling plan for each DER and the power purchase plan for the VPP control center. The intraday energy deviation coordination optimization model uses an iterative algorithm to obtain the user's real-time power consumption strategy that minimizes the energy deviation. The step S3 includes the following sub-steps: S3-1. Initialize the forecast value of the total load power on the user side and the forecast power generation of each DER on the day before. S3-2. In the day-ahead phase, the distributed algorithm of ADE nested Gurobi solver is used to solve the energy supply side scheduling optimization model. The VPP control center transmits the initialized VPP internal power purchase price to each DER in the lower layer. The DERs in the lower layer use Gurobi solver to optimize their energy sales plan according to their own power generation forecast, and obtain the power output result of the lower layer. And transmitted to the upper VPP control center, where is the total transaction power between PV and VPP control center at time t, is the total transaction power between WT and VPP control center at time t, is the total transaction power between GT and VPP control center at time t; VPP control center uses ADE algorithm to calculate its benefit value C VPP , and update the electricity price, and pass it to the lower layer for optimization; until the ADE algorithm converges, the optimal VPP energy acquisition plan and DERs energy scheduling plan are obtained, and the energy acquisition plan is passed to the intraday stage for optimization; S3-3. In the intraday stage, real-time demand response and cooperative game are considered moment by moment. At moment t, the VPP control center determines whether the load is within the allowable power deviation range of ±2% based on the day-ahead power purchase plan and the current real-time load. If it is within the allowable power deviation range, there is no power deviation penalty at the current moment, no operation is performed, and the process proceeds to the next moment. If it exceeds the allowable power deviation range of ±2%, an iterative algorithm is used to solve the energy deviation collaborative optimization model based on the cooperative game to obtain the user's real-time power consumption strategy that minimizes the energy deviation.

2. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 1 is characterized in that: The unified modeling of the energy storage system equipped with each DER is as follows: in, is the operating cost of the energy storage device; is the charging and discharging power of the energy storage device at time t; is the amount of electricity stored in ES at time t.

3. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 2 is characterized in that: The constraints for the unified modeling of the energy storage systems equipped by each DER are: in, Respectively represent the maximum power of charging and discharging; are 0-1 variables, representing the charging state and discharging state of ESS at time t.

4. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 3 is characterized in that: In the energy supply side dispatch optimization model of the day-ahead stage, the photovoltaic power generation unit takes its own income F PV Maximization is the optimization goal, and its objective function is: Among them, λ t is the electricity price within the VPP; is the total transaction power between PV and VPP control center at time t; Equip photovoltaic power stations with energy storage charging and discharging power; The operating cost of the ES equipped for PV; Predicting power generation for PV; The wind turbine generator system uses its own income F WT Maximization is the optimization goal, and its objective function is: in, is the total transaction power between WT and VPP control center at time t; Equip WT with energy storage for charging and discharging power; The operating cost of the ES equipped for WT; Predicting power generation for WT; Gas turbine with its own income F GT Maximization is the optimization goal, and its objective function is: Where, is the total transaction power between GT and VPP control center at time t; is the output power of GT at time t in the day-ahead power production scheduling; Equip GT with energy storage charging and discharging power; The running costs of the ES equipped for the GT; is the operating cost of GT, k GT is the cost coefficient of GT.

5. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 4 is characterized in that: In the energy supply side dispatch optimization model of the day-ahead stage, the upper-level leader VPP control center uses the VPP power purchase cost C VPP Minimization is the optimization goal, and its objective function is: Among them, C VPP The cost of purchasing electricity for the VPP control center; The amount of electricity agreed upon between the VPP control center and the power grid company; The agreed electricity price; The predicted power of the user side load at time t; is the total output power of DER inside VPP at time t.

6. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 5 is characterized in that: The energy supply-side dispatch optimization model in the day-ahead phase sets relevant electricity price constraints to adjust the electricity price within a specific range, while the average 24-hour electricity price is not lower than the market electricity price. The specific constraints are as follows: Among them, λ mk is the average market electricity price; The upper and lower limits of the VPP internal electricity price; is the electricity price within 24 hours λ t The average value of .

7. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 1 is characterized in that: The intra-day phase energy deviation coordination optimization model includes a cooperative alliance composed of multiple users. Each user in the cooperative alliance uses electricity cost. Minimization is the optimization goal, and the constraints of the objective function include: The user can reduce the load control range constraint settings as follows: To ensure the completion of daily transferable load tasks, the relevant capacity constraints are set as follows: Among them, E i is the total amount of transferable load that user i needs to complete within 24 hours; Δt is the intraday decision step, and Δt is 30 minutes; When making a decision at time t within a day, user i needs to make the following judgments: If the judgment result of (I) is true, it means that user i's transferable load task for the day has been completed, and the transferable load at time t and the subsequent time of the day is zero; if the judgment result of (II) is true, it means that the transferable load at time t and the subsequent time of the day needs to run at full power to complete the transferable load task for the day; in other cases, user i can freely control the start and stop of the transferable load as needed.

8. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 7 is characterized in that: The cooperative alliance uses electricity cost Minimum is the optimization goal, and its objective function is: in, The penalty cost caused by the power deviation of the daily load, is the predicted power of user-side load at time t, The amount of electricity agreed upon between the VPP control center and the power grid company; is the total output power of DER inside VPP at time t.

9. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 1 is characterized in that: The intra-day energy deviation coordination optimization model uses the Shapley value method to allocate incentive subsidies to users, and the allocation rules are as follows: Where N is the set of participants in the alliance, represented as {user 1, user 2, ..., user i}; is the benefit obtained by alliance member e; φ(S) is the weight of the benefit shared by alliance members; S\{i} is the set after member i is excluded from set S; e(S) is the characteristic function of the alliance; v(S) is the total benefit of the alliance; x i is the income of member i before participating in the cooperative game; S is the set of different alliance combinations.

10. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 9 is characterized in that: The step S3-2 includes the following sub-steps: S3-2-1. Initialize the simulation parameters of the energy supply side scheduling optimization model in the day-ahead phase, which include the operating cost coefficient β of the energy storage system; the charging and discharging power coefficient ρ of the ESS. ch , ρ dis ;Optimize time step Δt; ESS capacity Maximum power of ESS charging and discharging GT cost coefficient k GT ; The average electricity purchase price provided by the market to DERs λ mk ;The maximum number of iterations K of the ADE algorithm max ; Population size n size ; The dimension n of each individual in the population dim ; Search interval for target solution The scaling factor F of population variation and the probability of crossover and recombination CR; S3-2-2. Initialize population a, i.e. randomly generate n populations within the VPP internal electricity price control constraints. size The energy purchase price is used as the initial population, and the number of iterations is initialized to K = 0; S3-2-3: The VPP internal energy purchase price in population a is transmitted to each DER in the lower layer. Each DER uses the Gurobi solver according to the electricity price to optimize its power output. Upload the optimization results to the upper-level leader VPP control center; S3-2-4. The VPP control center calculates the benefit values of each individual in population a according to the lower-layer optimization results, and extracts the maximum benefit value E1 in this round; S3-2-5. Mutate and crossover population a to obtain a new population b; S3-2-6. Transmit the internal energy acquisition price of VPP in population b to each DER in the lower layer. The DERs call the Gurobi solver again to optimize their objective functions, and transmit the optimization results to the upper layer. The VPP control center calculates the benefit values of each individual in population b according to the returned lower-layer optimization results, and extracts the maximum benefit value E2 in this round; S3-2-7. Selection operation. If E2 > E1, then a = b and E1 = E2. If E2 < E1, remain unchanged; S3-2-8, judgment operation, if the maximum number of iterations K=K is reached max , then output the optimization result, otherwise jump to S3-2-5.

11. The two-stage game operation optimization method for a virtual power plant considering real-time demand response according to claim 10 is characterized in that: The step S3-3 includes the following sub-steps: S3-3-1. Initialize the simulation parameters of the intra-day energy deviation coordination optimization model in step S1, the simulation parameters including the restriction coefficient ε of each user's electricity load reduction cut,i ; User load reduction discomfort coefficient μ cut,i ; Uncomfort coefficient μ of user load transfer tr,i ; Rated power P for transferable load operation rated ; Unit penalty cost for energy deviation Unit penalty cost for energy deviation S3-3-2, at time t, compare the current real-time load with the power purchase plan at the same time in the day-ahead power purchase plan to determine whether the load deviation is within the allowable power deviation range of ±2%, that is, If it is within the allowable power deviation range, there is no power deviation penalty at the current moment, no operation is performed, and the process goes to S3-3-10; otherwise, the process goes to S3-3-3; At S3-3-3, at time t, the load deviation on the user side exceeds the allowable range. The VPP control center issues a load power adjustment call. Each user adjusts the flexible load demand to the maximum extent within their load adjustment range, and obtains the power consumption strategy of each user. S3-3-4. Calculate each user's incentive allowance based on the Shapley value method And calculate the total energy cost of each user S3-3-5. Determine whether the power deviation exceeds the allowable deviation range. If it exceeds the range, it means that the maximum load response at time t cannot completely eliminate the energy deviation. The game stops and the power consumption strategy of each user is output. Go to S3-3-10; otherwise go to S3-3-6; S3-3-6. Within the load adjustment range, each user's response load power is halved to obtain the power consumption strategy of each user. S3-3-7. Determine whether the current round of power deviation exceeds the allowable deviation range. If it exceeds the range, output the power usage strategy for each user. Go to S3-3-10; otherwise go to S3-3-8; S3-3-8. According to the electricity usage strategy Calculate incentive allowances for each user And calculate the total energy cost of each user Compare each user's and like Use renew Using policy B, user i Update policy A for user i otherwise User i's electricity usage strategy remains unchanged; S3-3-9. Determine whether all users meet If satisfied, all users determine their own DR strategies, the game ends, and the user power consumption strategy is output. Go to S3-3-10; Otherwise, go to S3-3-6; S3-3-10. The collaborative optimization of the power deviation at time t ends, and wait to enter time t+1.

Images