Photovoltaic micro-grid cooperative scheduling method based on multiple subjects

Abstract

The invention belongs to the technical field of photovoltaic micro-grids, and particularly relates to a photovoltaic micro-grid cooperative scheduling method based on multiple subjects. In order to solve the problem that EV user benefits, EV dynamic characteristics and the satisfaction degree of users to charging cost are not fully considered, the method is based on a prediction error and user sensitivity-based master-slave game scheduling strategy, firstly, a single-leader and multi-follower master-slave game model is established, a photovoltaic micro-grid operator serves as a leader in the master-slave game, and the photovoltaic micro-grid operator serves as the leader in the master-slave game; an EV aggregator and a vehicle owner serve as followers in a master-slave game, a game model is solved through mixed integer linear programming, an EV charging load prediction error and user sensitivity are considered during solving, a real-time dynamic electricity price correction mechanism is established by using an EV charging load historical prediction error and user sensitivity, a game feasible region is effectively optimized, and a real-time dynamic electricity price correction mechanism is established. Therefore, the EV user actively responds to the electricity price issued by the operator and carries out the charging behavior in order, thereby improving the photovoltaic local consumption.

Description

Technical Field

[0001] The present invention belongs to the technical field of photovoltaic microgrids, and in particular relates to a photovoltaic microgrid coordinated scheduling method based on multiple agents. Background Art

[0002] While related technologies are constantly innovating and developing, the effective absorption of distributed photovoltaic power in microgrids and the problem of reverse overload have become urgent problems to be solved. Electric vehicles can reasonably allocate the power supply and demand of EVs due to their dispatchability, becoming an effective way to solve this problem. Photovoltaic microgrids can use photovoltaic output to charge EVs and make full use of renewable energy. At the same time, EV travel demand can further promote the development of photovoltaic microgrids and form a benign vehicle-grid interaction. Through the synergy of the two, more efficient and environmentally friendly energy utilization methods can be achieved. The document "Multi-time scale random optimization scheduling of electric vehicle charging stations considering demand response" established an optimization model with the maximum benefit of charging stations, but did not fully consider the interests of EV users; the document "Dual-layer optimization scheduling of electric vehicle charging and discharging with the best interests of source, grid and load" established an EV intraday optimal scheduling model to minimize the peak-to-valley difference of the power grid load, but failed to reflect the dynamic characteristics of EVs; the document "Research on the economic absorption of wind and solar power based on game theory" established a responsive multi-objective optimization scheduling model for load fluctuations and profit maximization of EV access to the power grid, but did not consider user satisfaction with charging costs. Patent 202410535274.4 uses a proximal strategy optimization algorithm to train a multi-agent reinforcement learning model, combines the current state quantity to determine the action space of each agent, and optimizes the scheduling of multi-energy coupled microgrids based on the action space. However, the performance of the proximal strategy optimization algorithm depends largely on the adjustment of hyperparameters, and the parameter adjustment process may be more complicated, and there is a disadvantage of low sample data utilization efficiency. Patent 202310745847.1 proposes a source-grid-load-storage coordinated scheduling solution method based on an improved genetic algorithm, but it does not solve the problem that the genetic algorithm has many control variables, resulting in poor correlation of data results and inconsistency with actual results.

[0003] In summary, the integration of EV and PV microgrid for coordinated planning and scheduling is of great significance to the orderly charging of EV and the consumption of renewable energy, and has important application value for the optimization of microgrid operation. Summary of the invention

[0004] In view of the problem that the interests of EV users, the dynamic characteristics of EVs and the user's satisfaction with the charging cost are not fully considered, the present invention provides a photovoltaic microgrid collaborative scheduling method based on multiple agents. The method is based on a master-slave game scheduling strategy based on prediction error and user sensitivity. First, a single-leader and multiple-follower master-slave game model is established. The photovoltaic microgrid operator serves as the leader in the master-slave game, and the EV aggregator and the car owner serve as the followers in the master-slave game. The game model is solved by mixed integer linear programming, and the time meter and EV charging load prediction error and user sensitivity are solved to improve the solution efficiency. The EV charging load historical prediction error and user sensitivity are used to establish a real-time dynamic price correction mechanism. The real-time dynamic price correction mechanism can effectively optimize the game feasible domain, so that EV users actively respond to the electricity price released by the operator and charge in an orderly manner, thereby improving the local photovoltaic consumption.

[0005] To achieve the above object, the present invention adopts the following technical solutions:

[0006] EV charging load is highly volatile and difficult to accurately predict. Therefore, the deep learning method based on empirical mode decomposition (EMD) can achieve good results in the field of EV charging load prediction, but it does not solve the problem of multiple models and complex modeling. The use of historical high-frequency modes to predict future EV high-frequency modes has limited effect, and the model does not learn high-frequency modes with large fluctuations very well. At the same time, it does not consider the classification of load mode fluctuation complexity and the impact of meteorological factors on user charging behavior. It has certain practical significance to mine the fluctuation characteristics of each mode after decomposing the EV charging load based on frequency domain features, determine the high and low frequency classification standards, find a suitable prediction model, and learn the characteristics of high-frequency mode fluctuations caused by meteorological factors.

[0007] The present invention provides a photovoltaic microgrid coordinated dispatching method based on multiple agents, comprising the following steps:

[0008] By clarifying the transaction model between photovoltaic microgrid operators and EV aggregators, the relationship between multiple subjects in the coordinated dispatch of photovoltaic microgrids is characterized based on the Stackelberg game, and a photovoltaic microgrid coordinated dispatch model based on non-cooperative game is constructed to obtain the optimal broadcast electricity price and EV charging strategy.

[0009] Furthermore, in the PV microgrid collaborative dispatching model based on non-cooperative game, the PV microgrid operator is in a dominant position, publishing broadcast electricity price information based on its own PV power generation and contracted electricity; EV aggregators and users are in a follower position, formulating the most favorable charging strategy based on the broadcast electricity price information, thereby launching a game between broadcast electricity prices and charging strategies.

[0010] Furthermore, the game between the broadcast electricity price and the charging strategy includes the following stages:

[0011] (1) The broadcasting electricity price is set by the operator t , EV aggregators receive broadcast electricity prices c t After that, formulate the corresponding optimal charging strategy p it and feed it back to the operator;

[0012] (2) After receiving the optimal charging strategy from the EV aggregator, the operator adjusts the electricity price according to its own interests and generates a new broadcast electricity price. Regarding the new broadcast electricity price EV aggregators redefine optimal charging strategy and feed it back to the operator;

[0013] (3) The operator continues to update the broadcast electricity price and sends it to the EV aggregator until the game reaches a state of equilibrium when neither party can gain more benefits by changing their strategies. At this point, the best strategy for both parties is the game equilibrium point.

[0014] Furthermore, the game equilibrium point The solution process is:

[0015] (1) The KKT condition is introduced to eliminate the optimization problem of the lower-level EV charging cost, the master-slave game problem is transformed into a quadratic programming problem, and the PV microgrid coordinated scheduling model based on non-cooperative game is linearized;

[0016] (2) Optimize the upper and lower limits of the feasible domain of broadcast electricity prices based on a real-time dynamic correction mechanism;

[0017] (3) Use CPLEX tools to solve the optimal broadcast electricity price and the optimal charging strategy.

[0018] Furthermore, the real-time dynamic correction mechanism is based on the prediction error and the EV user sensitivity matrix, specifically:

[0019] Calculate the historical EV charging load forecast error, output the average forecast error one week before the dispatch day, and optimize the upper and lower limits of the feasible domain of broadcast electricity prices based on the forecast error and the electricity price increase and decrease coefficient; the microgrid operator calculates the user sensitivity matrix on the dispatch day, and further optimizes the upper and lower limits of the feasible domain of broadcast electricity prices based on the EV user's response to the broadcast electricity price.

[0020] Furthermore, the prediction error of the historical EV charging load is calculated, and the average prediction error of the week before the dispatch day is output. The upper and lower limits of the broadcast electricity price are optimized according to the prediction error and the electricity price increase and decrease coefficient, specifically:

[0021] Calculate the EV charging load forecast error one week before the dispatch date, average the historical forecast errors at the same time, and calculate the average forecast error e at time T q , the operator uses the average prediction error e at time T q , optimize broadcast electricity prices t The upper and lower limits of the feasible region are calculated as follows:

[0022]

[0023]

[0024]

[0025]

[0026] Among them, e q is the historical forecast error, is the EV charging load forecast value, is the historical value of EV charging load, The broadcast electricity price floor optimized for historical forecast errors, Broadcast electricity price cap optimized for historical forecast error, e qf is the average error of historical forecasts, Indicates raising the lower limit of broadcast electricity price, Indicates lowering the lower limit of broadcast electricity price, is the upper limit of the broadcast electricity price. qf <e q <e qf When the original broadcast electricity price ceiling is maintained, q is the electricity price increase coefficient, and p is the electricity price decrease coefficient.

[0027] Furthermore, the microgrid operator calculates the user sensitivity matrix on the dispatch day, and further optimizes the upper and lower limits of the feasible domain of the broadcast electricity price according to the sensitivity of EV users to the broadcast electricity price, specifically:

[0028] According to the EV charging load forecast results, the sensitivity matrix of EV user response electricity price is determined, and the initial broadcast electricity price c is further adjusted according to the user sensitivity. t The upper and lower limits of the feasible region, the calculation formula of the EV user electricity price sensitivity matrix E is as follows:

[0029]

[0030] Among them, E(i,i) is the sensitivity matrix of charging demand to electricity price, c(i) is the electricity price, b is the slope, and a is the intercept of the demand function on the axis;

[0031] The relationship between charging demand and electricity price can be expressed as a linear relationship:

[0032] p(i)=a-bc(i)

[0033] Among them, p(i) is the charging demand, which is determined by the microgrid operator through the load forecast results.

[0034] The sensitivity of charging demand to electricity prices is as follows:

[0035]

[0036] The two formulas can be used to obtain this calculation formula.

[0037] Compared with the prior art, the present invention has the following advantages:

[0038] The multi-agent-based photovoltaic microgrid collaborative scheduling method proposed in the present invention can realize system optimization scheduling more efficiently, maximize the interests of photovoltaic microgrid operators and electric vehicle users, increase the efforts to guide the effective scheduling of electric vehicles, and improve the local consumption of new energy power generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is the process of solving the game equilibrium point.

[0040] Figure 2 This is the EV charging load prediction result.

[0041] Figure 3 This is the photovoltaic power prediction result.

[0042] Figure 4 To compare the broadcast electricity price boundaries before and after optimization. DETAILED DESCRIPTION

[0043] In order to further illustrate the technical solution of the present invention, the present invention is further described below through embodiments.

[0044] Example 1

[0045] A photovoltaic microgrid coordinated dispatching method based on multiple agents in this embodiment includes the following steps:

[0046] By clarifying the transaction model between photovoltaic microgrid operators and EV aggregators, the relationship between multiple subjects in the coordinated dispatch of photovoltaic microgrids is characterized based on the Stackelberg game, and a photovoltaic microgrid coordinated dispatch model based on non-cooperative game is constructed to obtain the optimal broadcast electricity price and EV charging strategy.

[0047] In the PV microgrid collaborative dispatching model based on non-cooperative game, the PV microgrid operator is in a dominant position, publishing broadcast electricity price information based on its own PV power generation and contracted electricity; EV aggregators and users are in a follower position, formulating the most favorable charging strategy based on the broadcast electricity price information, thus launching a game between broadcast electricity prices and charging strategies.

[0048] The game between broadcast electricity prices and charging strategies includes the following stages:

[0049] (1) The broadcasting electricity price is set by the operator t , EV aggregators receive broadcast electricity prices c t After that, formulate the corresponding optimal charging strategy p it and feed it back to the operator;

[0050] (2) After receiving the optimal charging strategy from the EV aggregator, the operator adjusts the electricity price according to its own interests and generates a new broadcast electricity price. Regarding the new broadcast electricity price EV aggregators redefine optimal charging strategy and feed it back to the operator;

[0051] (3) The operator continues to update the broadcast electricity price and sends it to the EV aggregator until the game reaches a state of equilibrium when neither party can gain more benefits by changing their strategies. At this point, the best strategy for both parties is the game equilibrium point.

[0052] In this process, in addition to establishing a game model, the difficulty lies in how to determine the game equilibrium point. Since operators need to consider the response of EV users to prices, c t With p it All are decision variables. The game problem posed by operators and EV users is neither linear nor convex. How to solve this problem is the key to determining the equilibrium point of the game.

[0053] Furthermore, the game equilibrium point of this embodiment is The solution process is (as Figure 1 shown):

[0054] (1) The KKT condition is introduced to eliminate the optimization problem of the lower-level EV charging cost, the master-slave game problem is transformed into a quadratic programming problem, and the PV microgrid coordinated scheduling model based on non-cooperative game is linearized;

[0055] (2) Optimize the upper and lower limits of the feasible domain of broadcast electricity prices based on a real-time dynamic correction mechanism;

[0056] (3) Use CPLEX tools to solve the optimal broadcast electricity price and the optimal charging strategy to guide EV users to charge in an orderly manner and absorb local photovoltaic output.

[0057] Figure 1 In the equation (2-1)-(2-7), the objective function and constraints are:

[0058]

[0059] Among them, c t represents the broadcast electricity price during period t; E t represents the contracted electricity volume in the day-ahead market during period t; They represent the purchase and sale of electricity by the operator in the real-time market during period t; the charging amount of the i-th EV during period t is p it ; is the electricity purchase price of the operator in the real-time market during period t; is the electricity price sold by the operator in the real-time market during period t; i is the weight of the satisfaction of the i-th car owner in time period t; v i represents the satisfaction function of the i-th car owner.

[0060]

[0061] in, is the electricity purchase price of the operator in the real-time market during period t; is the electricity price sold by the operator in the real-time market during period t; represents the electricity purchase price in the day-ahead market during period t; c t represents the broadcast electricity price during period t; Broadcast electricity price cap, Broadcast electricity price floor.

[0062]

[0063]

[0064]

[0065]

[0066] Among them, E t represents the contracted electricity volume in the day-ahead market during period t; M refers to the ideal state of charge; They represent the electricity purchased and sold by operators in the real-time market during period t; θ t is a Boolean variable representing the energy trading status of the operator during period t; the charging amount of the i-th EV during period t is p it ; The real-time photovoltaic power generation in period t is v i represents the satisfaction function of the i-th car owner; β i Represents the owner priority factor.

[0067]

[0068] Among them, r i is the ideal charging capacity of the i-th EV, S iM is the ideal state of charge of the ith EV, S i0 is the initial state of charge of the i-th EV when it starts charging; is the battery capacity of the i-th EV.

[0069] Introducing dual variables

[0070] The duality theory performs KKT processing on the linear programming equations (2-8)-(2-10) to obtain the constraint equations (4-4)-(4-7):

[0071]

[0072] Among them, p it is the charging power of the ith electric vehicle in period t, {α it} corresponds to the optimal electric vehicle charging strategy; c t represents the broadcast electricity price during period t;

[0073]

[0074] in, is the initial power of the i-th electric car;

[0075]

[0076] in, represents the maximum charging power of the i-th electric vehicle in time period t; p it is the charging power of the ith electric vehicle in period t.

[0077]

[0078]

[0079]

[0080]

[0081] Among them, μ i , and are the dual variables of constraints (2-8) and (2-9), βi represents the owner priority factor, r i is the ideal charging capacity of the i-th EV, w and x represent the vertical and horizontal coordinate values ​​of the linear programming formula respectively.

[0082] Constraints (4-6) and (4-7) are complementary relaxation conditional formulas, introducing Boolean variables

[0083] Linearize constraints (4-6) and (4-7) into constraints (4-8)-(4-11);

[0084]

[0085]

[0086]

[0087]

[0088] Among them, Z is a sufficiently large number (set to the maximum transaction volume).

[0089] Substituting constraints (2-8)-(2-10) into the objective function yields the dual problem (4-12):

[0090]

[0091] in, is the battery capacity of the i-th EV, is the initial charge of the i-th electric car, and the dual formula λ becomes π.

[0092] The real-time dynamic correction mechanism is based on the prediction error and the EV user sensitivity matrix. The correction process is:

[0093] Calculate the historical EV charging load forecast error, output the average forecast error one week before the dispatch day, and optimize the upper and lower limits of the feasible domain of broadcast electricity prices based on the forecast error and the electricity price increase and decrease coefficient; the microgrid operator calculates the user sensitivity matrix on the dispatch day, and further optimizes the upper and lower limits of the feasible domain of broadcast electricity prices based on the EV user's response to the broadcast electricity price.

[0094] The specific correction steps are as follows:

[0095] 1) Calculate the EV charging load forecast error one week before the dispatch date, average the historical forecast errors at the same time, and calculate the average forecast error e at time T q , the operator uses the average prediction error e at time T q , optimize broadcast electricity prices t The upper and lower limits of the feasible region are calculated as follows:

[0096]

[0097]

[0098]

[0099]

[0100] Among them, e q is the historical forecast error, is the EV charging load forecast value, is the historical value of EV charging load, The broadcast electricity price floor optimized for historical forecast errors, Broadcast electricity price cap optimized for historical forecast error, e qf is the average error of historical forecasts, Indicates raising the lower limit of broadcast electricity price, Indicates lowering the lower limit of broadcast electricity price, is the upper limit of the broadcast electricity price. qf <e q <e qf When the original broadcast electricity price ceiling is maintained, q is the electricity price increase coefficient, and p is the electricity price decrease coefficient.

[0101] 2) According to the EV charging load forecast results, determine the sensitivity matrix of EV user response electricity price, and further adjust the initial broadcast electricity price c according to user sensitivity t Feasible region upper and lower limits, EV user electricity price sensitivity matrix E The calculation formula is as follows:

[0102]

[0103] Where c(i) is the electricity price, b is the slope, and a is the intercept of the demand function on the axis.

[0104] Example 2

[0105] The Stackelberg game scheduling model is established with the goal of maximizing the profit of photovoltaic microgrid operators and minimizing the charging cost of EVs. In the day-ahead market, operators need to predict the local photovoltaic power and EV charging load, and the input is the historical photovoltaic power and EV charging load of a certain place from January to July 2023. The microgrid operator calculates the predicted data (EV charging load and photovoltaic power generation power prediction results are as follows: Figure 2 and 3 The day-ahead power purchase contract is formulated using the power purchase quantity and the day-ahead power purchase cost as outputs.

[0106] In the real-time market, operators formulate broadcast electricity prices based on photovoltaic output and EV charging demand, and EV aggregators adjust their charging strategies based on the broadcast electricity prices. The two restrict each other and are connected through the broadcast electricity prices, thus forming a master-slave game with operators as leaders and EV users as followers.

[0107] The steps of real-time dynamic correction of the feasible domain of broadcast electricity price based on prediction error and EV user sensitivity matrix are as follows:

[0108] 1) Calculate the EV charging load forecast error for the week before the 180th day, average the historical forecast errors at the same time, and calculate T The average prediction error at each moment is used to optimize the upper and lower limits of the broadcast electricity price issued by the operator. The following table 1 shows the prediction error results.

[0109] Table 1 Prediction error

[0110]

[0111]

[0112] 2) The operator provides the e q Optimizing broadcast electricity price using calculation formula t The upper and lower limits of the feasible region.

[0113] 3) According to the EV charging load forecast results, the sensitivity matrix of EV user response electricity price is determined according to the following formula, and the initial broadcast electricity price c is further adjusted according to the user sensitivity t The upper and lower limits of the feasible domain. EV user electricity price sensitivity matrix E:

[0114]

[0115] Calculate the broadcast electricity price c based on the time-of-use electricity price in the region t Upper and lower limits As shown in Table 2 and Figure 4 shown.

[0116] Table 2. Boundary values ​​of broadcast electricity prices before and after optimization

[0117]

[0118]

[0119] In order to evaluate the performance of the model proposed in the present invention, a master-slave game collaborative scheduling model (basic collaborative scheduling model) that does not consider prediction error and user sensitivity is introduced for comparative simulation. In the master-slave game basic collaborative scheduling model, the boundary of the broadcast electricity price is determined according to the time-of-use electricity price disclosed by the government. The simulation results are shown in Tables 3 and 4.

[0120] Table 3 Model performance

[0121]

[0122] Table 4 Scheduling scheme

[0123]

[0124] The above shows and describes the main features and advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and scope of the equivalent elements of the claims are included in the present invention.

[0125] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A photovoltaic microgrid coordinated dispatching method based on multiple agents, characterized in that: The following steps are involved: By clarifying the transaction model between photovoltaic microgrid operators and EV aggregators, the relationship between multiple subjects in the coordinated dispatch of photovoltaic microgrids is characterized based on the Stackelberg game, and a photovoltaic microgrid coordinated dispatch model based on non-cooperative game is constructed to obtain the optimal broadcast electricity price and EV charging strategy.

2. A photovoltaic microgrid coordinated dispatching method based on multiple agents according to claim 1, characterized in that: In the PV microgrid collaborative dispatching model based on non-cooperative game, the PV microgrid operator is in a dominant position, publishing broadcast electricity price information based on its own PV power generation and contracted electricity; EV aggregators and users are in a follower position, formulating the most favorable charging strategy based on the broadcast electricity price information, thus launching a game between broadcast electricity price and charging strategy.

3. A photovoltaic microgrid coordinated dispatching method based on multiple agents according to claim 2, characterized in that: The game between the broadcast electricity price and the charging strategy includes the following stages: (1) The broadcasting electricity price is set by the operator t , EV aggregators receive broadcast electricity prices c t After that, formulate the corresponding optimal charging strategy p it and feed it back to the operator; (2) After receiving the optimal charging strategy from the EV aggregator, the operator adjusts the electricity price according to its own interests and generates a new broadcast electricity price. Regarding the new broadcast electricity price EV aggregators redefine optimal charging strategy and feed it back to the operator; (3) The operator continues to update the broadcast electricity price and sends it to the EV aggregator until the game reaches a state of equilibrium when neither party can gain more benefits by changing their strategies. At this point, the best strategy for both parties is the game equilibrium point.

4. A photovoltaic microgrid coordinated dispatching method based on multiple agents according to claim 3, characterized in that: The game equilibrium point The solution process is: (1) The KKT condition is introduced to eliminate the optimization problem of the lower-level EV charging cost, the master-slave game problem is transformed into a quadratic programming problem, and the PV microgrid coordinated scheduling model based on non-cooperative game is linearized; (2) Optimize the upper and lower limits of the feasible domain of broadcast electricity prices based on a real-time dynamic correction mechanism; (3) Use CPLEX tools to solve the optimal broadcast electricity price and the optimal charging strategy.

5. A photovoltaic microgrid coordinated dispatching method based on multiple agents according to claim 4, characterized in that: The real-time dynamic correction mechanism is based on the prediction error and the EV user sensitivity matrix, specifically: Calculate the historical EV charging load forecast error, output the average forecast error one week before the dispatch day, and optimize the upper and lower limits of the feasible domain of broadcast electricity prices based on the forecast error and the electricity price increase and decrease coefficient; the microgrid operator calculates the user sensitivity matrix on the dispatch day, and further optimizes the upper and lower limits of the feasible domain of broadcast electricity prices based on the EV user's response to the broadcast electricity price.

6. A photovoltaic microgrid coordinated dispatching method based on multiple agents according to claim 5, characterized in that: The prediction error of the historical EV charging load is calculated, and the average prediction error of the week before the dispatch date is output. The upper and lower limits of the broadcast electricity price are optimized according to the prediction error and the electricity price increase and decrease coefficient, specifically: Calculate the EV charging load forecast error one week before the dispatch date, average the historical forecast errors at the same time, and calculate the average forecast error at time T. The operator optimizes the broadcast electricity price c based on the average forecast error at time T. t The upper and lower limits of the feasible region are calculated as follows: Among them, e q is the historical forecast error, is the predicted value of EV charging load, P r EV is the historical value of EV charging load, The broadcast electricity price floor optimized for historical forecast errors, Broadcast electricity price cap optimized for historical forecast error, e qf is the average error of historical forecasts, Indicates raising the lower limit of broadcast electricity price, Indicates lowering the lower limit of broadcast electricity price, is the upper limit of the broadcast electricity price. qf <e q <e qf When the original broadcast electricity price ceiling is maintained, q is the electricity price increase coefficient, and p is the electricity price decrease coefficient.

7. A photovoltaic microgrid coordinated dispatching method based on multiple agents according to claim 6, characterized in that: The microgrid operator calculates the user sensitivity matrix on the dispatch day and further optimizes the upper and lower limits of the feasible domain of broadcast electricity prices according to the sensitivity of EV users to the broadcast electricity prices. Specifically, According to the EV charging load forecast results, the sensitivity matrix of EV user response electricity price is determined, and the initial broadcast electricity price c is further adjusted according to the user sensitivity. t The upper and lower limits of the feasible region, the calculation formula of the EV user electricity price sensitivity matrix E is as follows: Where c(i) is the electricity price, b is the slope, and a is the intercept of the demand function on the axis.

Images