Optimal day-ahead bidding strategy for virtual power plants considering demand response and frequency regulation performance changes

Abstract

A method for optimizing day-ahead bidding for virtual power plants (VPPs) that considers demand response and frequency regulation performance variations includes establishing the VPP's organizational structure and operating mechanism, modeling price-based demand response for distributed resources, modeling a VPP's secondary frequency regulation performance indicator function, and implementing a VPP day-ahead bidding constraint model. To address the uncertainties in wind power output and the increased demand for secondary frequency regulation in the power grid caused by wind turbines replacing conventional units, this method uses a profit-driven approach to encourage wind power and generalized energy storage to form VPPs, which then participate in the energy-frequency regulation market. This approach not only increases the number of VPP hours participating in the frequency regulation market and their total frequency regulation capacity, but also enhances the VPP's overall frequency regulation performance, providing higher-quality frequency regulation services for the system.

Description

Technical Field

[0001] The present invention relates to the field of power market, in particular to fields related to demand response and virtual power plants, and specifically to an optimal day-ahead bidding strategy for virtual power plants that takes into account changes in demand response and frequency regulation performance. Background Art

[0002] Wind power, as one of the major renewable energy sources, has experienced rapid growth. By the end of 2020, the cumulative installed capacity of grid-connected wind power had accounted for 12.8% of the nation's total installed capacity. However, the uncertainty of wind power output and the replacement of conventional units by wind turbines have increased the demand for secondary frequency regulation on the power grid, adversely impacting the safe and stable operation of the grid. Therefore, while wind power participates in the electricity energy market, it is also necessary to participate in the auxiliary frequency regulation market to alleviate the pressure on the power grid to regulate frequency. Frequency regulation refers to secondary frequency regulation; this invention does not involve primary frequency regulation.

[0003] Market-based electricity trading has played a positive role in improving power quality and promoting the integration of renewable energy. Some literature has explored the integration of wind power into the electricity market, introducing approaches such as demand response trading markets, peak-shaving rights trading, and supply-demand interactive market mechanisms to promote wind power integration. Advances in energy storage technology and the reduction in storage costs have made it possible to integrate energy storage systems into the power grid and provide services to the power system. Wind power can participate in system frequency regulation by controlling rotor speed and pitch angle. The rapid response of energy storage can offset the uncertainty of wind power, improve its frequency regulation performance, and increase frequency regulation benefits. In addition to traditional energy storage batteries, generalized energy storage, such as electric vehicles and air conditioning clusters, can also serve as energy storage units to provide services to wind power without the need for additional storage units. However, energy storage resources such as electric vehicles and air conditioning clusters are often widely distributed, have small capacity, and exhibit certain uncertainties, making them unsuitable for inclusion in the electricity market as standalone energy storage resources.

[0004] Virtual power plants (VPs) can aggregate multiple distributed resources and participate in the electricity market as a whole. Currently, some provinces and municipalities in my country have enacted policies and mechanisms for VPs to participate in the electricity market. Research has been conducted domestically and internationally on VPs that incorporate wind power. Through research, Chen W et al., in their paper "Bargaining game-based profit allocation of virtual powerplant in frequency regulation market considering battery cycle life," published in IEEE Transactions on Smart Grid, proposed an optimal bidding strategy for VPs in conjunction with wind farms in the frequency regulation ancillary services market. Wang Xing et al., in their paper "Game model for wind power and electric vehicles to form a virtual power plant participating in the electricity market," published in Automation of Power Systems, demonstrated that wind power companies and electric vehicle aggregators can achieve greater profits when participating in the electricity market through a VP partnership. However, none of these studies consider the impact of changes in frequency regulation performance indicators on bidding strategies. Furthermore, most literature does not consider the application of generalized energy storage in VPs.

[0005] Currently, no description or report of similar technology to the present invention has been found, and similar information at home and abroad has not been collected. Summary of the Invention

[0006] Based on the above background, this paper proposes a bidding strategy for a virtual power plant with wind and generalized energy storage participating in the energy-frequency regulation joint market, taking into account the impact of frequency regulation performance indicators. Stevens' law is used to model the response of electric vehicles and air conditioning clusters to the compensation electricity price, and a functional relationship is constructed between frequency regulation performance and the bid frequency regulation capacity of wind turbines, reserved frequency regulation capacity, and generalized energy storage frequency regulation capacity. With maximizing wind farm revenue as the optimization goal, a bidding strategy for the virtual power plant in the day-ahead market is formulated according to the electricity market trading rules. Finally, a numerical example is used to reflect the impact of frequency regulation performance on the bidding strategy, and the economic and effectiveness of the proposed strategy is confirmed.

[0007] The technical solutions of the present invention are as follows:

[0008] An optimal day-ahead bidding strategy for virtual power plants that considers demand response and frequency regulation performance variations mainly includes five parts: constructing the organizational structure and operating mechanism of a virtual power plant that includes wind power and generalized energy storage; using Stevens' law to model the response of electric vehicles and air conditioning clusters to compensation electricity prices; analyzing the factors affecting the frequency regulation performance of the virtual power plant and modeling the secondary frequency regulation performance index function of the virtual power plant; proposing an objective function and constraints for the day-ahead bidding of the virtual power plant based on the internal structure and operation mode of the virtual power plant; based on the proposed objective function and constraints, and introducing robust optimization to consider the uncertainty of wind power output, solving the resulting nonlinear mixed integer programming, and obtaining the optimal day-ahead bidding decision for the virtual power plant that considers demand response and frequency regulation performance variations. The specific steps are as follows:

[0009] 1. Virtual power plant organizational structure and operation mechanism

[0010] Organizational structure: The virtual power plant proposed in this invention aggregates three types of distributed resources: wind turbines, electric vehicles, and air conditioning clusters. Wind turbines are the main components of the virtual power plant, while electric vehicles and air conditioning clusters are flexible and fast-response resources that coordinate with wind turbines to accept scheduling and jointly participate in the electricity market. Figure 1 As shown in the figure, the virtual power plant as a whole appears to be "power-generating" to the outside world, and the distributed loads within the virtual power plant are all powered by wind power.

[0011] Operation mechanism: In the day-ahead stage, wind power, electric vehicles and air-conditioning clusters report the day-ahead predicted power and available power in each time period to the virtual power plant service center respectively. Based on the information reported by wind power and generalized energy storage and the clearing price forecast of the typical day of the electricity market, the virtual power plant service center formulates the optimal bidding strategy and declares the capacity to the electricity market. The dispatching center formulates the energy base point and frequency regulation capacity of the virtual power plant for the next day based on the declared capacity of the virtual power plant. In the real-time stage, wind power, electric vehicles and air-conditioning clusters jointly provide the energy base point and frequency regulation capacity required by the system. If the real-time energy base point or real-time frequency regulation capacity does not meet the system requirements, the missing capacity will be punished. The specific process is as follows: Figure 2 shown.

[0012] 2. Demand response model based on Stevens' law

[0013] a) Electric vehicle response model

[0014] In order to attract electric vehicles to participate in the scheduling and operation of virtual power plants, compensate for the uncertainty of wind power output, and improve the frequency regulation performance of wind power, the present invention proposes an electric vehicle charging and discharging induction mechanism based on time-of-use electricity prices and a variable subsidy mechanism to encourage electric vehicles to participate in frequency regulation services. The subsidy mechanism for participating in frequency regulation changes the subsidized electricity price for electric vehicles participating in frequency regulation, and dispatches appropriate electric vehicles to participate in frequency regulation in a time period of 15 minutes. For electric vehicles that do not participate in frequency regulation services, the charging and discharging induction mechanism is used to change the inherent charging and discharging habits of electric vehicles by changing the electricity price, and guide electric vehicles to charge and discharge in specific time periods. The present invention uses Stevens' law as the pricing basis to construct a response model for electric vehicles to frequency regulation services and charging and discharging behavior.

[0015] Assume that the compensation price is c a When , electric vehicles begin to accept virtual power plant dispatch, and electric vehicles participate in frequency regulation; the compensation price is c b When all electric vehicles participate in frequency regulation, the frequency regulation capacity of electric vehicles is the total grid-connected capacity, then the response rate of electric vehicles to frequency regulation service is Expressed as:

[0016]

[0017] Where: Frequency regulation compensation for electric vehicles, is the frequency modulation response coefficient of the electric vehicle, Tuning sensory index for electric vehicles.

[0018] Electric vehicle frequency regulation capacity during period t for:

[0019]

[0020] Where: is the responsiveness of electric vehicles participating in frequency modulation during period t, P EV,t is the total capacity of electric vehicles that can participate in frequency regulation during period t;

[0021] For electric vehicles that do not participate in frequency regulation, time-of-use electricity prices are used to induce electric vehicles to change their charging and discharging behaviors to maximize their profits. The response of these electric vehicles to charging and discharging also satisfies Stevens' law:

[0022]

[0023]

[0024] Where: c ch The electricity price for charging electric vehicles, R ch is the charge response rate, c dis is the electric vehicle discharge price, Rdis is the discharge response rate, c c0 and c c1 are the lowest and highest electricity prices for electric vehicles to participate in charging, k ch is the electric vehicle charging response coefficient, n ch Charging an electric car sensory index, c d0 and c d1 are the lowest and highest electricity prices for electric vehicles to participate in discharge, k dis is the discharge response coefficient of the electric vehicle, n dis It is the discharge sensory index of electric vehicles.

[0025] b) Air conditioning cluster response model

[0026] Assuming that the number of air conditioning clusters participating in the virtual power plant remains unchanged, the frequency regulation capacity of the air conditioning cluster is determined by the user's responsiveness, that is,

[0027]

[0028] Where: is the frequency modulation capacity of ACLs in period t, is the responsiveness of ACLs in frequency modulation during period t, P ACL,t is the operating power of ACLs in period t.

[0029]

[0030] Where, The subsidized electricity price for the air conditioner to participate in the frequency regulation service is is the frequency modulation response coefficient of the air conditioning cluster, The sensory index of the air conditioning cluster is FM, c m and c n They are respectively the minimum and maximum electricity prices for the air-conditioning cluster to participate in frequency regulation.

[0031] When the air conditioner starts operating, it will first enter the operating state until the room temperature reaches the preset lower limit of temperature fluctuation. If it does not accept virtual power plant dispatch, the air conditioner will operate in automatic start-stop mode, maintaining the room temperature within the preset temperature range. If the subsidized electricity price meets the user's psychological expectations and the air conditioner participates in the frequency modulation service and accepts virtual power plant dispatch, it will start and stop according to the virtual power plant's requirements, while the room temperature still remains within the acceptable fluctuation range. After the frequency modulation service ends, the air conditioner will first enter the operating state until the room temperature reaches the lower limit of the acceptable range, and then operate in automatic start-stop mode.

[0032] The automatic start-stop cycle for air conditioning clusters is typically less than half an hour. However, when the indoor and outdoor temperature difference is large, the start-stop cycle may even be less than 15 minutes. In such cases, to ensure user comfort, the air conditioning cluster will enter an automatic start-stop cycle regardless of the time-of-use electricity price. Therefore, a charge-discharge induction mechanism will not be introduced for air conditioning clusters.

[0033] 3. Frequency modulation performance index estimation function

[0034] In actual dispatching, the wind turbine first meets its bid volume in the energy market and operates based on the bid volume. Since the wind turbine also participates in the bidding in the frequency regulation market, during each trading period, the wind turbine also needs to adjust its output according to the AGC instructions issued by the dispatch center within the bid range of the frequency regulation capacity. When the real-time reserved frequency regulation capacity is less than the frequency regulation bid volume, it is very likely that the real-time output of the wind turbine will not be able to meet the AGC signal. Therefore, the frequency regulation performance comprehensive index S is related to the day-ahead frequency regulation bid volume. Real-time reserved frequency modulation capacity related.

[0035] When the fan output is insufficient, the fan frequency regulation performance will drop significantly, and may even fall below the PJM market requirements and be unacceptable.

[20] The electric vehicles and air conditioning clusters gathered in the virtual power plant can provide the frequency regulation requirements of the dispatching center by charging and discharging, changing the operating status, etc., reducing the frequency regulation deviation. At the same time, the rapid response capabilities of electric vehicles and air conditioning clusters enable these two generalized energy storages to compensate for the frequency regulation deviation caused by the insufficient adjustment rate of the fan, further improving the frequency regulation performance of the fan. Therefore, the comprehensive index of frequency regulation performance S is affected by the frequency regulation capacity of electric vehicles. Air conditioning cluster frequency modulation capacity impact.

[0036] In summary, the frequency modulation performance index S is affected by the above four variables, and the frequency modulation performance index estimation function should be a four-element function, which can be recorded as Since S will be used as a parameter in the final optimization model, the function will be fitted in the form of a polynomial for the convenience of calculation. The expression of the function can be obtained by fitting specific data.

[0037] 4. Objective function and constraints for day-ahead bidding of virtual power plants

[0038] a) Objective function

[0039] The purpose of this invention is to encourage wind power to participate in the frequency regulation market, combine wind power with generalized energy storage, give full play to the potential of distributed generalized energy storage, and provide high-performance frequency regulation services for the power grid. The goal is to maximize the benefits of wind farms:

[0040] C=C en +Cre +C EV +C ACL -C PUN (7)

[0041] The objective function consists of five parts: energy market revenue C en , FM market revenue C re 、Electric vehicle electricity sales revenue C EV , Air conditioning cluster electricity sales revenue C ACL and penalty cost C PUN .

[0042] Energy market revenue C en Determined by the bidding strategy and energy clearing price:

[0043]

[0044] Where: is the energy clearing price in period i, Bid capacity for the energy market during period i.

[0045] FM market revenue C re Divided into capacity gain C re,fc and adjust mileage earnings C re,fp :

[0046]

[0047] Where: is the clearing price of frequency regulation capacity in period i, is the price of frequency modulation mileage clearing in period i, S i is the frequency regulation performance index of period i, λ is the mileage benefit factor, Bid capacity for the frequency regulation market during period i.

[0048] Virtual power plants are price takers in the electricity market, and all bids for declared capacity will be awarded. In order to prevent virtual power plants from deliberately falsifying and disrupting the market order, a penalty cost C is introduced. PUN :

[0049]

[0050]

[0051]

[0052] Where: ρ en (·) and ρ re (·) are penalty functions corresponding to the energy market and frequency modulation market respectively, is the real-time energy base point of period i, The real-time frequency modulation capacity in time period i.

[0053] Electric vehicles are both consumers of wind power and providers of frequency regulation services. The revenue from electric vehicle electricity sales is C EV Profits from selling wind power to electric vehicles and frequency regulation subsidy costs It consists of two parts:

[0054]

[0055] Where: c ch,t 、c dis,t are the time-of-use electricity prices for electric vehicle charging and discharging during period t, P ch,t 、P dis,t are the charging and discharging powers of the electric vehicle during period t, is the compensation price for electric vehicles participating in frequency regulation during period t, is the frequency regulation capacity of electric vehicles in period t.

[0056] Air conditioning cluster electricity sales revenue C ACL :

[0057]

[0058] Where: The profit from selling wind power to air conditioning clusters is Subsidize the cost of frequency regulation for air conditioning clusters, c ACL The electricity price for wind power supply, The power consumption of the air conditioning cluster that does not participate in frequency regulation service. The electricity price for frequency regulation subsidy of air-conditioning cluster during period t.

[0059] b) Constraints

[0060] 1) Capacity Constraints

[0061] The day-ahead bid volume for virtual power plants is subject to the forecast output of wind power, the dispatch output of electric vehicles, and the dispatch output of air conditioning clusters:

[0062]

[0063]

[0064] Where: is the day-ahead predicted power of wind power in period i, P EV,i is the maximum output power of the electric vehicle in period i, P ACL,i is the maximum output power of the air-conditioning cluster in period i, Δt is the time interval of period t, and Δi is the time interval of period i.

[0065] The frequency modulation capacity and charging and discharging power of electric vehicles are subject to the maximum available power of electric vehicles:

[0066]

[0067] The frequency modulation capacity of the air conditioning cluster is subject to the maximum power constraint of the air conditioning cluster:

[0068]

[0069] Frequency regulation bidding capacity is subject to wind power rated capacity and energy market bidding capacity constraints:

[0070]

[0071]

[0072]

[0073] 2) Electric vehicle state of charge constraints

[0074] The SOC of a certain period is constrained by the SOC of the previous period and the energy change of the current period:

[0075] SOC k,t =SOC k,t-1 +ΔSOC k,t (twenty two)

[0076]

[0077] Where: SOC k,t is the average state of charge of k-type electric vehicles during period t, ΔSOC k,t is the change of SOC during period t, and σ is the charging and discharging efficiency of the electric vehicle.

[0078] Deep charging and discharging will shorten the battery life. SOC is constrained by the remaining energy of the battery:

[0079] SOC min ≤SOC k,t ≤SOC max (twenty four)

[0080] Where: SOC min and SOC max They are the lower and upper limits of the state of charge allowed for electric vehicle batteries.

[0081] The final state during the optimization period is constrained by the initial state:

[0082] SOC k,0 =SOC k,95 +ΔSOC k,96 =SOC k,96 (25)

[0083] 3) Air conditioning cluster room temperature constraints

[0084] The indoor temperature should be kept within the allowable range of the set temperature fluctuation:

[0085]

[0086] Where: The preset upper and lower limits of indoor temperature fluctuation.

[0087] 5. Robust optimization model and solution for virtual power plant day-ahead bidding

[0088] In the objective function, the decision variables are the day-ahead energy market bid amount, the day-ahead frequency regulation market bid amount, the next day electric vehicle frequency regulation compensation price, the next day electric vehicle charging price, the next day electric vehicle discharging price, and the next day air conditioning cluster frequency regulation compensation price. The uncertain variables are the real-time output of wind power and the actual market clearing price.

[0089] According to historical statistical data, the real-time output of wind turbines is generally less than the day-ahead power forecast, and the deviation is generally within 15%. Assuming that the real-time output fluctuation of wind power is within the range of [-15%, +5%], that is, Assuming that the fluctuation range of the market clearing price is ±10%, the present invention performs robust optimization based on the day-ahead power forecast and the clearing price on a typical day, obtains the optimal bidding strategy, and calculates the actual profit of wind power based on the real-time wind power output.

[0090] The proposed model is a nonlinear mixed integer programming. In the MATLAB R2021 environment, YALMIP+GUROBI is used to solve the model.

[0091] Compared with the prior art, the present invention has the following characteristics:

[0092] 1. Propose an optimal day-ahead bidding strategy for virtual power plants that considers demand response and frequency regulation performance variations;

[0093] 2. A novel virtual power plant organizational structure and operating mechanism is proposed. This virtual power plant structure incorporates generalized energy storage, such as wind turbines, electric vehicles, and air conditioning clusters. The proposed operating mechanism is innovative and can fully leverage the potential value of distributed generalized energy storage while facilitating the management and control of distributed resources.

[0094] 3. Demand response modeling is based on Stevens' Law, which reflects the relationship between stimulus intensity and sensory perception. Using this law better reflects the psychological changes of users of distributed resources, making the proposed strategy more practical.

[0095] 4. Establish a frequency regulation performance index prediction function. In the optimization model, the existing technology defines the frequency regulation performance index as a constant, while the present invention uses the frequency regulation performance index prediction function to dynamically evaluate the frequency regulation performance index. This change will encourage virtual power plants to participate in frequency regulation services, alleviate frequency regulation pressure, and provide better quality frequency regulation services. BRIEF DESCRIPTION OF THE DRAWINGS

[0096] Figure 1 This is the organizational structure diagram of the virtual power plant proposed in the present invention;

[0097] Figure 2 Schematic diagram of the virtual power plant operation mechanism proposed in the present invention;

[0098] Figure 3 It is a dynamic evaluation diagram of the performance index of the frequency modulation performance comprehensive index fitting function in a specific example of the present invention;

[0099] Figure 4 It is a bidding result diagram according to the prior art in a specific embodiment of the present invention;

[0100] Figure 5 It is a bidding result diagram according to the bidding strategy of the present invention in a specific example of the present invention. DETAILED DESCRIPTION

[0101] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented based on the technical solution of the present invention, provides a detailed implementation method and specific operating process, and further illustrates the present invention with reference to the embodiments and drawings. It should be noted that those skilled in the art may make various modifications and improvements without departing from the scope of the present invention, and all such modifications and improvements fall within the scope of protection of the present invention.

[0102] An optimal day-ahead bidding strategy for virtual power plants that considers demand response and frequency regulation performance variations mainly includes five parts: constructing the organizational structure and operating mechanism of a virtual power plant that includes wind power and generalized energy storage; using Stevens' law to model the response of electric vehicles and air conditioning clusters to compensation electricity prices; analyzing the factors affecting the frequency regulation performance of the virtual power plant and modeling the secondary frequency regulation performance index function of the virtual power plant; proposing an objective function and constraints for the day-ahead bidding of the virtual power plant based on the internal structure and operation mode of the virtual power plant; based on the proposed objective function and constraints, and introducing robust optimization to consider the uncertainty of wind power output, solving the resulting nonlinear mixed integer programming, and obtaining the optimal day-ahead bidding decision for the virtual power plant that considers demand response and frequency regulation performance variations. The specific steps are as follows:

[0103] 1. Virtual power plant organizational structure and operation mechanism

[0104] Organizational structure: The virtual power plant proposed in this invention aggregates three types of distributed resources: wind turbines, electric vehicles, and air conditioning clusters. Wind turbines are the main components of the virtual power plant, while electric vehicles and air conditioning clusters are flexible and fast-response resources that coordinate with wind turbines to accept scheduling and jointly participate in the electricity market. Figure 1 As shown in the figure, the virtual power plant as a whole appears to be "power-generating" to the outside world, and the distributed loads within the virtual power plant are all powered by wind power.

[0105] Operation mechanism: The present invention refers to the operation mechanism of the PJM electricity market in the United States. The energy market and the frequency regulation auxiliary market are jointly cleared, and the trading stage is divided into the day-ahead market and the intraday real-time market. In the day-ahead market, power generators submit bidding information on a time scale of 1 hour. After stopping bidding, the market enters the clearing stage. PJM determines the hourly unit combination, the day-ahead node marginal electricity price, and the demand for day-ahead planned reserve based on the unit combination analysis, and announces the day-ahead hourly power generation plan and node marginal cost. The real-time market is a supplement to the day-ahead market. It provides market members with an opportunity to adjust the dispatch plan. The real-time market is cleared every 5 minutes or 15 minutes according to the real-time node marginal electricity price under the actual grid operating conditions.

[0106] The frequency regulation market utilizes a performance-based revenue mechanism. The settlement price is the sum of the capacity price and the performance index price. Higher performance indexes indicate better frequency regulation services provided by frequency regulation resources, which translates to higher revenue for frequency regulation resources. In the real-time market, PJM clears the frequency regulation price every 5 minutes. All 5-minute frequency regulation prices within an hour are averaged to calculate the hourly frequency regulation price.

[0107] In the day-ahead stage, wind power, electric vehicles, and air-conditioning clusters report the day-ahead predicted power and available power in each time period to the virtual power plant service center respectively. Based on the information reported by wind power and generalized energy storage and the clearing price forecast of a typical day in the electricity market, the virtual power plant service center formulates the optimal bidding strategy and declares capacity to the electricity market. The dispatching center formulates the energy base point and frequency regulation capacity of the virtual power plant for the next day based on the declared capacity of the virtual power plant. In the real-time stage, wind power, electric vehicles, and air-conditioning clusters jointly provide the energy base point and frequency regulation capacity required by the system. If the real-time energy base point or real-time frequency regulation capacity does not meet the system requirements, the missing capacity will be penalized. The specific process is as follows: Figure 2 shown.

[0108] Since the capacities of wind power and generalized energy storage are relatively small, they serve only as price receivers in the electricity market, submitting only capacity during the bidding phase, and all declared capacity will be adopted. This paper focuses on the proposed bidding strategy for virtual power plants in the day-ahead market, determining the bidding capacity of virtual power plants in the energy market and frequency regulation market with the goal of maximizing their total revenue. At the same time, to incentivize wind power companies to achieve the most accurate forecasts possible and reduce the error between the bid capacity in the day-ahead market and the actual output, a deviation penalty cost is introduced to curb participants' false reporting behavior.

[0109] 2. Demand response model based on Stevens' law

[0110] a) Electric vehicle response model

[0111] This paper studies the charging and discharging behavior of electric vehicle groups during the operation of a virtual power plant. Considering the charging and discharging behavior of each electric vehicle would make the solution process extremely cumbersome and unnecessary. Therefore, this paper uses the k-means clustering algorithm to cluster electric vehicles within the jurisdiction of the virtual power plant, selecting the time when electric vehicles are connected to the grid and the state of charge (SOC) when connected to the grid as characteristic parameters. To facilitate subsequent calculations, the number of clusters in the data set is set between [2, 10]. The Calinski-Harabasz (CH) index size is calculated for each cluster number, and the cluster number with the maximum CH is taken as the k value for clustering.

[0112] After clustering, the geometric center of each category is selected as the behavior rule of all electric vehicles in that category, that is, it is assumed that all electric vehicles in each category operate with the grid connection and off-grid time represented by the geometric center and the SOC when connected to the grid.

[0113] In order to attract electric vehicles to participate in the scheduling and operation of virtual power plants, compensate for the uncertainty of wind power output, and improve the frequency regulation performance of wind power, the present invention proposes an electric vehicle charging and discharging induction mechanism based on time-of-use electricity prices and a variable subsidy mechanism to encourage electric vehicles to participate in frequency regulation services. The subsidy mechanism for participating in frequency regulation changes the subsidized electricity price for electric vehicles participating in frequency regulation, and dispatches appropriate electric vehicles to participate in frequency regulation in a time period of 15 minutes. For electric vehicles that do not participate in frequency regulation services, the charging and discharging induction mechanism is used to change the inherent charging and discharging habits of electric vehicles by changing the electricity price, and guide electric vehicles to charge and discharge in specific time periods. The present invention uses Stevens' law as the pricing basis to construct a response model for electric vehicles to frequency regulation services and charging and discharging behavior.

[0114] Assume that the compensation price is c a When , electric vehicles begin to accept virtual power plant dispatch, and electric vehicles participate in frequency regulation; the compensation price is c bWhen all electric vehicles participate in frequency regulation, the frequency regulation capacity of electric vehicles is the total grid-connected capacity, then the response rate of electric vehicles to frequency regulation service is Expressed as:

[0115]

[0116] Where: Frequency regulation compensation for electric vehicles, is the frequency modulation response coefficient of the electric vehicle, Tuning sensory index for electric vehicles.

[0117] Electric vehicle frequency regulation capacity during period t for:

[0118]

[0119] Where: is the responsiveness of electric vehicles participating in frequency modulation during period t, P EV,t is the total capacity of electric vehicles that can participate in frequency regulation during period t;

[0120] For electric vehicles that do not participate in frequency regulation, time-of-use electricity prices are used to induce electric vehicles to change their charging and discharging behaviors to maximize their profits. The response of these electric vehicles to charging and discharging also satisfies Stevens' law:

[0121]

[0122]

[0123] Where: c ch The electricity price for charging electric vehicles, R ch is the charge response rate, c dis is the electric vehicle discharge price, R dis is the discharge response rate, c c0 and c c1 are the lowest and highest electricity prices for electric vehicles to participate in charging, k ch is the electric vehicle charging response coefficient, n ch Charging an electric car sensory index, c d0 and c d1 are the lowest and highest electricity prices for electric vehicles to participate in discharge, k dis is the discharge response coefficient of the electric vehicle, n dis It is the discharge sensory index of electric vehicles.

[0124] b) Air conditioning cluster response model

[0125] Air conditioning loads are modeled using a method based on second-order equivalent thermal parameters. Air conditioners generally operate near a preset temperature with automatic start-stop. When the air conditioner is started, and the indoor temperature reaches the preset temperature, it automatically stops and enters an idle state. Due to the temperature difference between indoors and outdoors, the indoor temperature will gradually approach the outdoor temperature over time. When the difference between the indoor temperature and the preset temperature exceeds the allowable range, the air conditioner automatically starts and enters the operating state, thus maintaining the indoor temperature near the preset temperature. The change pattern of indoor temperature over time when the air conditioner is on and off is as follows:

[0126] T in,t+1 =T out -(T out -T in,t )e -Δt / RC s=0 (5)

[0127]

[0128] Where: s represents the operating state of the air conditioner. When s = 0, the air conditioner is in idle state. When s = 1, the air conditioner is in working state. in,t is the indoor temperature during period t, T in,t+1 is the indoor temperature during period t+1, T out is the outdoor temperature, assuming that the outdoor temperature remains unchanged, Δt is the time interval of each period, that is, 15 minutes, R is the equivalent thermal resistance, C is the equivalent heat capacity, η is the air conditioner energy efficiency ratio, P0 is the air conditioner rated power, and A is the thermal conductivity.

[0129] Similar to the electric vehicle response model, assuming that the number of air conditioning clusters participating in the virtual power plant remains unchanged, the frequency regulation capacity of the air conditioning cluster is determined by the user's responsiveness, that is,

[0130]

[0131] Where: is the frequency modulation capacity of ACLs in period t, is the responsiveness of ACLs in frequency modulation during period t, P ACL,t is the operating power of ACLs in period t.

[0132]

[0133] Where, The subsidized electricity price for the air conditioner to participate in the frequency regulation service is is the frequency modulation response coefficient of the air conditioning cluster, The sensory index of the air conditioning cluster is FM, c m and c n They are respectively the minimum and maximum electricity prices for the air-conditioning cluster to participate in frequency regulation.

[0134] When the air conditioner starts operating, it will first enter the operating state until the room temperature reaches the preset lower limit of temperature fluctuation. If it does not accept virtual power plant dispatch, the air conditioner will operate in automatic start-stop mode, maintaining the room temperature within the preset temperature range. If the subsidized electricity price meets the user's psychological expectations and the air conditioner participates in the frequency modulation service and accepts virtual power plant dispatch, it will start and stop according to the virtual power plant's requirements, while the room temperature still remains within the acceptable fluctuation range. After the frequency modulation service ends, the air conditioner will first enter the operating state until the room temperature reaches the lower limit of the acceptable range, and then operate in automatic start-stop mode.

[0135] The automatic start-stop cycle for air conditioning clusters is typically less than half an hour. However, when the indoor and outdoor temperature difference is large, the start-stop cycle may even be less than 15 minutes. In such cases, to ensure user comfort, the air conditioning cluster will enter an automatic start-stop cycle regardless of the time-of-use electricity price. Therefore, a charge-discharge induction mechanism will not be introduced for air conditioning clusters.

[0136] 3. Frequency modulation performance index estimation function

[0137] In actual dispatching, the wind turbine first meets its bid volume in the energy market and operates based on the bid volume. Since the wind turbine also participates in the bidding in the frequency regulation market, during each trading period, the wind turbine also needs to adjust its output according to the AGC instructions issued by the dispatch center within the bid range of the frequency regulation capacity. When the real-time reserved frequency regulation capacity is less than the frequency regulation bid volume, it is very likely that the real-time output of the wind turbine will not be able to meet the AGC signal. Therefore, the frequency regulation performance comprehensive index S is related to the day-ahead frequency regulation bid volume. Real-time reserved frequency modulation capacity related.

[0138] When the fan output is insufficient, the fan frequency regulation performance will drop significantly, and may even fall below the PJM market requirements and be unacceptable.

[20] The electric vehicles and air conditioning clusters gathered in the virtual power plant can provide the frequency regulation requirements of the dispatching center by charging and discharging, changing the operating status, etc., reducing the frequency regulation deviation. At the same time, the rapid response capabilities of electric vehicles and air conditioning clusters enable these two generalized energy storages to compensate for the frequency regulation deviation caused by the insufficient adjustment rate of the fan, further improving the frequency regulation performance of the fan. Therefore, the comprehensive index of frequency regulation performance S is affected by the frequency regulation capacity of electric vehicles. Air conditioning cluster frequency modulation capacity impact.

[0139] In summary, the frequency modulation performance index S is affected by the above four variables, and the frequency modulation performance index estimation function should be a four-element function, which can be recorded as Since S will be used as a parameter in the final optimization model, the function will be fitted in the form of a polynomial for the convenience of calculation. The expression of the function can be obtained by fitting specific data.

[0140] The valuation function should contain something like and At the same time, in order to reduce the nonlinearity of the valuation function and facilitate the solution of the optimization model, a polynomial is used for fitting, and finally the function with the best fitting effect is obtained. This function has 10 items, and and There are 4 related items in total, and the remaining items are all items with 2 or less times, as shown below:

[0141]

[0142] The goodness of fit is 0.966, and the fitting effect is good. The average deviation between the estimated data and the experimental data of the fitting function is 1.36%, which is small and meets the engineering requirements. The function can be used to optimize the model. The fitting effect diagram is shown in the figure. Figure 3 shown.

[0143] 4. Objective function and constraints for day-ahead bidding of virtual power plants

[0144] a) Objective function

[0145] The purpose of this invention is to encourage wind power to participate in the frequency regulation market, combine wind power with generalized energy storage, give full play to the potential of distributed generalized energy storage, and provide high-performance frequency regulation services for the power grid. The goal is to maximize the benefits of wind farms:

[0146] C=C en +C re +C EV +C ACL -C PUN (10)

[0147] The objective function consists of five parts: energy market revenue C en , FM market revenue C re 、Electric vehicle electricity sales revenue C EV , Air conditioning cluster electricity sales revenue C ACL and penalty cost C PUN .

[0148] Energy market revenue C en Determined by the bidding strategy and energy clearing price:

[0149]

[0150] Where: is the energy clearing price in period i, Bid capacity for the energy market during period i.

[0151] FM market revenue C re Divided into capacity gain C re,fc and adjust mileage earnings C re,fp :

[0152]

[0153] Where: is the clearing price of frequency regulation capacity in period i, is the price of frequency modulation mileage clearing in period i, S i is the frequency regulation performance index of period i, λ is the mileage benefit factor, Bid capacity for the frequency regulation market during period i.

[0154] Virtual power plants are price takers in the electricity market, and all bids for declared capacity will be awarded. In order to prevent virtual power plants from deliberately falsifying and disrupting the market order, a penalty cost C is introduced. PUN :

[0155]

[0156]

[0157]

[0158] Where: ρ en (·) and ρ re (·) are penalty functions corresponding to the energy market and frequency modulation market respectively, is the real-time energy base point of period i, The real-time frequency modulation capacity in time period i.

[0159] Electric vehicles are both consumers of wind power and providers of frequency regulation services. The revenue from electric vehicle electricity sales is C EV Profits from selling wind power to electric vehicles and frequency regulation subsidy costs It consists of two parts:

[0160]

[0161] Where: c ch,t 、c dis,t are the time-of-use electricity prices for electric vehicle charging and discharging during period t, P ch,t 、P dis,t are the charging and discharging powers of the electric vehicle during period t, is the compensation price for electric vehicles participating in frequency regulation during period t, is the frequency regulation capacity of electric vehicles in period t.

[0162] Air conditioning cluster electricity sales revenue C ACL :

[0163]

[0164] Where: The profit from selling wind power to air conditioning clusters is Subsidize the cost of frequency regulation for air conditioning clusters, c ACL The electricity price for wind power supply, The power consumption of the air conditioning cluster that does not participate in frequency regulation service. The electricity price for frequency regulation subsidy of air-conditioning cluster during period t.

[0165] b) Constraints

[0166] 1) Capacity Constraints

[0167] The day-ahead bid volume for virtual power plants is subject to the forecast output of wind power, the dispatch output of electric vehicles, and the dispatch output of air conditioning clusters:

[0168]

[0169]

[0170] Where: is the day-ahead predicted power of wind power in period i, P EV,i is the maximum output power of the electric vehicle in period i, P ACL ,i is the maximum output power of the air conditioning cluster in period i, Δt is the time interval of period t, and Δi is the time interval of period i.

[0171] The frequency modulation capacity and charging and discharging power of electric vehicles are subject to the maximum available power of electric vehicles:

[0172]

[0173] The frequency modulation capacity of the air conditioning cluster is subject to the maximum power constraint of the air conditioning cluster:

[0174]

[0175] Frequency regulation bidding capacity is subject to wind power rated capacity and energy market bidding capacity constraints:

[0176]

[0177]

[0178]

[0179] 2) Electric vehicle state of charge constraints

[0180] The SOC of a certain period is constrained by the SOC of the previous period and the energy change of the current period:

[0181] SOC k,t =SOC k,t-1 +ΔSOC k,t (25)

[0182]

[0183] Where: SOC k,t is the average state of charge of k-type electric vehicles during period t, ΔSOC k,t is the change of SOC during period t, and σ is the charging and discharging efficiency of the electric vehicle.

[0184] Deep charging and discharging will shorten the battery life. SOC is constrained by the remaining energy of the battery:

[0185] SOC min ≤SOC k,t ≤SOC max (27)

[0186] Where: SOC min and SOC max They are the lower and upper limits of the state of charge allowed for electric vehicle batteries.

[0187] The final state during the optimization period is constrained by the initial state:

[0188] SOC k,0 =SOC k,95 +ΔSOC k,96 =SOC k,96 (28)

[0189] 3) Air conditioning cluster room temperature constraints

[0190] The indoor temperature should be kept within the allowable range of the set temperature fluctuation:

[0191]

[0192] Where: The preset upper and lower limits of indoor temperature fluctuation.

[0193] 5. Robust optimization model and solution for virtual power plant day-ahead bidding

[0194] In the objective function, the decision variables are the day-ahead energy market bid amount, the day-ahead frequency regulation market bid amount, the next day electric vehicle frequency regulation compensation price, the next day electric vehicle charging price, the next day electric vehicle discharging price, and the next day air conditioning cluster frequency regulation compensation price. The uncertain variables are the real-time output of wind power and the actual market clearing price.

[0195] According to historical statistical data, the real-time output of wind turbines is generally less than the day-ahead power forecast, and the deviation is generally within 15%. Assuming that the real-time output fluctuation of wind power is within the range of [-15%, +5%], that is, Assuming that the fluctuation range of the market clearing price is ±10%, the present invention performs robust optimization based on the day-ahead power forecast and the clearing price on a typical day, obtains the optimal bidding strategy, and calculates the actual profit of wind power based on the real-time wind power output.

[0196] The proposed model is a nonlinear mixed integer programming. In the MATLAB R2021 environment, YALMIP+GUROBI is used to solve the model.

[0197] The above technical solution of the present invention is further described below with reference to a specific example.

[0198] This specific example involves a wind farm with a rated installed capacity of 200 MW, 200 electric vehicles of the same model, and a cluster of 300 air conditioners of the same model. Assume that the wind farm's upward and downward frequency modulations are each 20% of the rated capacity. Due to limitations on wind turbine ramp rates and to avoid ramping events, the standard wind turbine modulation rate is 3% of the rated capacity per minute. The wind turbine control error follows a normal distribution with a mean of 0 and a variance of 0.000013. The unit's permissible response delay is 1 minute.

[0199] The day-ahead power forecast and real-time output of wind turbines in a certain region on September 1, 2021, were obtained from the PJM official website. Based on the installed capacity of the region, the obtained data was converted into the day-ahead power forecast and real-time output of a wind farm with a rated capacity of 200 MW; and the corresponding energy-frequency regulation market price was obtained for post-calculation of revenue.

[0200] This specific example designed two scenarios for comparative experiments: Scenario 1, the wind farm participates in the electricity market alone, and the comprehensive frequency regulation performance index is calculated based on historical frequency regulation conditions and is a constant; Scenario 2, wind power, electric vehicles, and air conditioning clusters together form a virtual power plant and participate in the electricity market in the form of a virtual power plant. The comprehensive frequency regulation performance index is predicted according to the estimation function required above and is a variable.

[0201] The bidding strategy of wind farm in scenario 1 is as follows: Figure 4 As shown in Table 1, in this scenario, the wind farm participates in the frequency regulation market for 15 hours, with a total bid capacity of 1,693 MW in the energy market and 980 MW in the frequency regulation market. In this scenario, the wind power frequency regulation performance index is 0.82. Market revenue is shown in Table 1, with the majority of revenue coming from the energy market.

[0202] The bidding strategy of virtual power plant in scenario 2 is as follows: Figure 5 As shown in Figure 1, in this scenario, the virtual power plant participates in the frequency regulation market for 20 hours, with a total bid capacity of 1,569 MW in the energy market and 1,312 MW in the frequency regulation market. In this scenario, since frequency regulation performance is roughly predictable and controllable, virtual power plants will participate more frequently in the frequency regulation market, providing more frequency regulation services to the system. To maximize profits, wind farms will actively encourage generalized energy storage to participate in frequency regulation, resulting in more efficient utilization of generalized energy storage. Due to the increase in energy storage frequency regulation capacity, the average frequency regulation performance index rises to 0.96, significantly improving frequency regulation performance. Table 1 shows the wind farm's revenue. Compared with Scenario 1, energy market revenue decreases by 7.70%, frequency regulation market revenue increases by 48.45%, and total net wind power revenue increases by 10.16%. The increase in wind farm revenue primarily comes from the frequency regulation market. Improved frequency regulation performance allows wind power to gain more economic benefits from the frequency regulation market, which in turn drives an increase in frequency regulation capacity. These two factors mutually reinforce each other, significantly increasing frequency regulation revenue.

[0203] Table 1 Comparison of wind power benefits under different scenarios

[0204]

Claims

1. An optimal day-ahead bidding method for virtual power plants considering demand response and frequency regulation performance changes, characterized by: include: Build the organizational structure and operating mechanism of a virtual power plant that includes wind power and generalized energy storage; Stevens' law is used to model the response of electric vehicles and air conditioning clusters to the compensation electricity price; Analyze the factors that affect the frequency regulation performance of virtual power plants and model the secondary frequency regulation performance index function of virtual power plants; Based on the internal structure and operation mode of virtual power plants, the objective function and constraints of day-ahead bidding for virtual power plants are proposed. Based on the proposed objective function and constraints, robust optimization is introduced to consider the uncertainty of wind power output. The resulting nonlinear mixed integer programming is solved to obtain the optimal day-ahead bidding decision for the virtual power plant, which takes into account the changes in demand response and frequency regulation performance. The modeling of the virtual power plant secondary frequency regulation performance index function includes: the frequency regulation performance index S affected by four variables, denoted as is the day-ahead frequency regulation bid amount, To reserve frequency modulation capacity in real time, Frequency regulation capacity for electric vehicles, Frequency modulation capacity for air conditioning clusters; The objective function and constraints of the virtual power plant day-ahead bidding include: a) Objective function With the goal of maximizing wind farm profits: C=C en +C re +C EV +C ACL -C PUN (7) Where: C en is the energy market income, C re For FM market revenue, C EV The income from selling electricity for electric vehicles, C ACL The electricity sales revenue of the air conditioning cluster, C PUN For penalty costs; Among them, the energy market income C en Determined by the bidding strategy and energy clearing price: Where: is the energy clearing price in period i, Bid capacity for the energy market during period i; FM market revenue C re Divided into capacity gain C re,fc and adjust mileage earnings C re,fp : Where: is the clearing price of frequency regulation capacity in period i, is the price of frequency modulation mileage clearing in period i, S i is the frequency regulation performance index of period i, λ is the mileage benefit factor, Bid capacity for the frequency regulation market during period i; Virtual power plants are price takers in the electricity market, and all bids for declared capacity will be awarded. In order to prevent virtual power plants from deliberately falsifying and disrupting the market order, a penalty cost C is introduced. PUN : Where: ρ en (·) and ρ re (·) are penalty functions corresponding to the energy market and frequency modulation market respectively, is the real-time energy base point of period i, is the real-time frequency modulation capacity during period i; Electric vehicles are both consumers of wind power and providers of frequency regulation services. The revenue from electric vehicle electricity sales is C EV Profits from selling wind power to electric vehicles and frequency regulation subsidy costs It consists of two parts: Where: c ch,t 、c dis,t are the time-of-use electricity prices for electric vehicle charging and discharging during period t, P ch,t 、P dis,t are the charging and discharging powers of the electric vehicle during period t, is the compensation price for electric vehicles participating in frequency regulation during period t, is the frequency regulation capacity of electric vehicles during period t; Air conditioning cluster electricity sales revenue C ACL : Where: The profit from selling wind power to air conditioning clusters is Subsidize the cost of frequency regulation for air conditioning clusters, c ACL The electricity price for wind power supply, The power consumption of the air conditioning cluster that does not participate in frequency regulation service. Subsidy electricity price for frequency regulation of air-conditioning cluster during period t; b) Constraints 1) Capacity Constraints The day-ahead bid volume for virtual power plants is subject to constraints such as wind power forecast output, electric vehicle dispatch output, and air conditioning cluster dispatch output: Where: is the day-ahead predicted power of wind power in period i, P EV,i is the maximum output power of the electric vehicle in period i, P ACL,i is the maximum output power of the air conditioning cluster in period i, Δt is the time interval of period t, Δi is the time interval of period i, The frequency modulation capacity and charging and discharging power of electric vehicles are subject to the maximum available power of electric vehicles: The frequency modulation capacity of the air conditioning cluster is subject to the maximum power constraint of the air conditioning cluster: Frequency regulation bidding capacity is subject to wind power rated capacity and energy market bidding capacity constraints: 2) Electric vehicle state of charge constraints The SOC of a certain period is constrained by the SOC of the previous period and the energy change of the current period: SOC k,t =SOC k,t-1 +ΔSOC k,t (22) Where: SOC k,t is the average state of charge of type k electric vehicles during period t, ΔSOC k,t is the change of SOC in period t, σ is the charging and discharging efficiency of the electric vehicle; Deep charging and discharging will shorten the battery life. SOC is constrained by the remaining energy of the battery: SOC min ≤SOC k,t ≤SOC max (24) Where: SOC min and SOC max They are the lower and upper limits of the state of charge allowed for electric vehicle batteries; The final state during the optimization period is constrained by the initial state: SOC k,0 =SOC k,95 +ΔSOC k,96 =SOC k,96 (25) 3) Air conditioning cluster room temperature constraints The indoor temperature should be kept within the allowable range of the set temperature fluctuation: Where: The preset upper and lower limits of indoor temperature fluctuation.

2. The optimal day-ahead bidding method for a virtual power plant considering demand response and frequency regulation performance changes according to claim 1 is characterized in that: The organizational structure of the virtual power plant includes three types of distributed resources: wind turbines, electric vehicles, and air conditioning clusters. The wind turbines serve as the primary components of the virtual power plant, while the electric vehicles and air conditioning clusters serve as flexibly dispatchable, fast-response resources that coordinate with the wind turbines to participate in the electricity market. The virtual power plant as a whole presents itself as a power generation facility, while the internal distributed loads are all powered by wind power. The operating mechanism is as follows: in the day-ahead stage: wind power, electric vehicles and air-conditioning clusters report the day-ahead predicted power and available power in each time period to the virtual power plant service center respectively; based on the information reported by wind power and generalized energy storage and the clearing price forecast of the typical day of the electricity market, the virtual power plant service center formulates the optimal bidding strategy and declares the capacity to the electricity market; the dispatching center formulates the energy base point and frequency regulation capacity of the virtual power plant for the next day based on the declared capacity of the virtual power plant; in the real-time stage: wind power, electric vehicles and air-conditioning clusters jointly provide the energy base point and frequency regulation capacity required by the system. If the real-time energy base point or real-time frequency regulation capacity does not meet the system requirements, the missing capacity will be punished.

3. The optimal day-ahead bidding method for a virtual power plant considering demand response and frequency regulation performance changes according to claim 1 is characterized in that: The modeling of the response to the compensation electricity price includes: a) Electric vehicle response model Assume that the compensation price is c a When , electric vehicles begin to accept virtual power plant dispatch, and electric vehicles participate in frequency regulation; the compensation price is c b When all electric vehicles participate in frequency regulation, the frequency regulation capacity of electric vehicles is the total grid-connected capacity, then the response rate of electric vehicles to frequency regulation service is Expressed as: Where: Frequency regulation compensation for electric vehicles, is the frequency modulation response coefficient of the electric vehicle, Tuning sensory index for electric vehicles; Electric vehicle frequency regulation capacity during period t for: Where: is the responsiveness of electric vehicles participating in frequency modulation during period t, P EV,t is the total capacity of electric vehicles that can participate in frequency regulation during period t; For electric vehicles that do not participate in frequency regulation, time-of-use electricity prices are used to induce electric vehicles to change their charging and discharging behaviors to maximize their profits. The response of these electric vehicles to charging and discharging also satisfies Stevens' law: Where: c ch The electricity price for charging electric vehicles, R ch is the charge response rate, c dis is the electric vehicle discharge price, R dis is the discharge response rate, c c0 and c c1 are the lowest and highest electricity prices for electric vehicles to participate in charging, k ch is the electric vehicle charging response coefficient, n ch Charging an electric car sensory index, c d0 and c d1 are the lowest and highest electricity prices for electric vehicles to participate in discharge, k dis is the discharge response coefficient of the electric vehicle, n dis It is the electric vehicle discharge sensory index; b) Air conditioning cluster response model Assuming that the number of air conditioning clusters participating in the virtual power plant remains unchanged, the frequency regulation capacity of the air conditioning cluster is determined by the user's responsiveness, that is, Where: is the frequency modulation capacity of ACLs in period t, is the responsiveness of ACLs in frequency modulation during period t, P ACL,t is the operating power of ACLs during period t; Where, The subsidized electricity price for the air conditioner to participate in the frequency regulation service is is the frequency modulation response coefficient of the air conditioning cluster, The sensory index of the air conditioning cluster is FM, c m and c n are the minimum and maximum electricity prices for the air conditioning cluster to participate in frequency regulation; When the air conditioner starts running, it will first enter the working state until the room temperature reaches the lower limit of the preset temperature fluctuation. If it does not accept the virtual power plant dispatch, the air conditioner will follow the automatic start and stop operation mode to keep the room temperature fluctuating within the preset temperature range. If the subsidized electricity price meets the user's psychological expectations, the air conditioner will participate in the frequency modulation service and accept the virtual power plant dispatch. In this case, the air conditioner will start and stop according to the virtual power plant's requirements, and the room temperature will still be kept within the allowed fluctuation range. After the frequency modulation service is finished, the air conditioner will first enter the working state until the room temperature reaches the lower limit of the allowable range, and then operate in the automatic start-stop mode; The automatic start-stop cycle of the air-conditioning cluster is generally less than half an hour, and when the temperature difference between indoor and outdoor is too large, the start-stop cycle may even be less than 15 minutes; at this time, in order to ensure the comfort of users, regardless of the time-of-use electricity price, the air-conditioning cluster will enter the automatic start-stop cycle; no charging and discharging induction mechanism will be introduced for the air-conditioning cluster.

4. The optimal day-ahead bidding method for a virtual power plant considering demand response and frequency regulation performance changes according to claim 1 is characterized in that: include: A robust optimization model for the day-ahead bidding strategy is established, and the nonlinear mixed integer model is solved using yalmip+gurobi. Specifically, the following are performed: In the objective function, the decision variables are the day-ahead energy market bid amount, the day-ahead frequency regulation market bid amount, the next day electric vehicle frequency regulation compensation price, the next day electric vehicle charging price, the next day electric vehicle discharging price, and the next day air conditioning cluster frequency regulation compensation price. The uncertain variables are the real-time output of wind power and the actual market clearing price. According to historical statistical data, the real-time output of wind turbines is generally less than the day-ahead power forecast, and the deviation is generally within 15%. Assuming that the real-time output fluctuation of wind power is within the range of [-15%, +5%], that is, Assuming that the fluctuation range of the market clearing price is ±10%, robust optimization is performed based on the day-ahead power forecast and the clearing price on a typical day to obtain the optimal bidding strategy, and the actual profit of wind power is calculated based on the real-time wind power output.

Images