A method for mining demand response potential of electric vehicle fast charging stations

Abstract

The present invention proposes a method for mining the demand response potential of electric vehicle fast charging stations. Considering that private electric vehicle fast charging station operators are not sensitive to traffic network information (such as electric vehicle travel demand, etc.), in order to avoid possible loss of charging station benefits and waste of demand response potential, the interaction process between charging station operators and the environment is modeled as a Markov chain, and the SAC (soft actor critic) algorithm is used to obtain its optimal strategy. At the same time, in order to avoid the demand response information output by the power grid causing the charging load to violate the power grid security constraints, the CPO (constrained policy optimization) algorithm is used to obtain the optimal strategy of the power grid. This method deeply mines the demand response potential of charging stations and is expected to provide new ideas for electric vehicles to participate in power grid scheduling.

Description

Technical Field

[0001] The invention relates to a method for mining demand response potential of an electric vehicle fast charging station, and belongs to the technical field of power systems. Background Art

[0002] The global trend of carbon reduction has promoted the widespread popularization of environmentally friendly electric vehicles. In order to meet the travel needs of electric vehicles, charging stations have been widely built and their number has increased significantly. As a result, the interdependence between the distribution network and the transportation network has been significantly enhanced.

[0003] There has been an increasing number of studies devoted to the coordinated dispatch of coupled power and transportation networks. Within the existing power-transportation coupled system dispatch framework, the current method of adjusting charging loads is to use marginal prices to set the charging fees of charging stations. This method manages the charging loads of charging stations with varying marginal prices by capturing the congestion state of the distribution network. However, the marginal price-based regulation method is based on the fully developed electricity spot market, which is not yet fully established. In addition, the deterministic marginal price has limited flexibility in managing the charging load of charging stations and ignores the profitability of charging stations, which may lead to a decrease in the interest of profit-seeking capital in investing and operating charging stations. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a method for mining the demand response potential of electric vehicle fast charging stations. By mining the demand response potential of fast charging stations, the power grid operation efficiency and charging facility utilization rate are improved, the charging cost of electric vehicles is reduced, and the popularization and sustainable development of electric vehicles are promoted.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for mining the demand response potential of an electric vehicle fast charging station, the specific steps are as follows:

[0007] In the first step, the distribution network, electric vehicle fast charging stations, and electric vehicle users in the transportation network are regarded as different stakeholders. Since the distribution network and charging station operators are partially unknown to the external environment, their decision-making process is modeled as a Markov chain.

[0008] In the second step, the distribution network operator obtains the distribution network operation parameters by solving the distribution network Distflow model and broadcasts the formulated demand response information to all electric vehicle charging station operators;

[0009] Solve the distribution network Distflow model to obtain operating parameters and output demand response information, including:

[0010] Consider having Represents a radial topology distribution network, where represents the set of lines in the distribution network g, and Represents the bus set in the distribution network g. The operating parameters of the distribution network are obtained by solving the following linear Disflow model;

[0011]

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019] Where P g,ij and Q g,ij represents the active / reactive power flow of line (i, j) in the distribution network g; is an auxiliary variable. The representative node j is the root node of the distribution network g; r g,ij and x g,ij are the resistance and reactance of line (i, j) respectively; U g,j , and Indicates the voltage amplitude, and its upper and lower limits; and Represents the total power load, conventional power load, and charging load; and Active / reactive power output and power factor of photovoltaic power generation;

[0020] Constraints (1)-(2) ensure the balance of active power and reactive power at each node in the distribution network, constraint (3) represents the relationship between voltage drop and active / reactive power on the line, and constraints (4)-(5) consider the power factor of photovoltaic power generation. At the same time, the equipment constraints are imposed on the active and reactive power output of the photovoltaic power. Constraint (6) represents the randomness of the photovoltaic active output and obeys a certain probability distribution. Constraint (7) summarizes the node power load, where the charging load It is determined by solving the traffic assignment model. Constraint (8) limits the range of voltage amplitude;

[0021] After obtaining the charging load of the EV fast charging station, the distribution network operator solves the linearized Distflow model to obtain detailed operation data. After obtaining the distribution network operation details, the charging load is adjusted through the generated demand response information, which is then sent to the EV fast charging station operator;

[0022]

[0023]

[0024]

[0025]

[0026]

[0027]

[0028]

[0029] Where g, k and i represent the indexes of distribution network, electric vehicle fast charging station and distribution network line respectively; G is the set of distribution network, K is the set of electric vehicle fast charging stations; C is the total cost of distribution network operator, including operating cost and demand response cost. represents the operating cost of the distribution network g, including the electricity charges for photovoltaic power generation cut-off and transformers, C DR represents the demand response cost for distribution network operators; represents the upper and lower demand response limits imposed by the distribution network operator on the billable load EV fast charging station k; represents the charging load k of the electric vehicle fast charging station; π e and π pv They represent the demand response price, the price of electricity purchased from the upper grid, and the penalty for underutilizing the active power of photovoltaic power;

[0030] Constraint (9) is the objective function of the distribution network operator, including the distribution network operating cost and demand response fee, constraints (10)-(12) are the methods for calculating the demand response cost and operating cost, and constraints (13)-(15) show the feasible area of ​​the demand response information formulated by the distribution network operator;

[0031] In the third step, after observing the historical and current distribution network electricity prices and demand response information, the charging station operator outputs the charging price of each charging station;

[0032] The fourth step is to obtain the final result of the user game by solving the traffic distribution model that takes into account user equilibrium, and return the charging load to the distribution network and charging stations;

[0033] Step 5: Iterate steps 2 to 4 repeatedly until the reward curve converges.

[0034] As a further improvement of the present invention, the operation constraints and profit objectives of the charging station operator in the objective function of the distribution network operator in step 2 include:

[0035] Considering the optimal pricing strategy for electric vehicle charging stations, the electricity selling price of the distribution network operator is regarded as the input of this process, and the charging price of the electric vehicle fast charging station is regarded as the output. Subsequently, the charging price is broadcast to all electric vehicle users. The electric vehicle fast charging station will use environmental information, including historical charging load and electricity price, to determine a reasonable charging price, while maximizing the revenue:

[0036]

[0037]

[0038]

[0039]

[0040] In the formula, H k represents the profit of electric vehicle fast charging station; k ,λ min ,λ max represents the charging price and its upper and lower limits; γ g,k is an auxiliary variable that specifies the connection relationship between the distribution network nodes and the EV fast charging stations;

[0041] Constraint (16) is the objective function of the EV fast charging station operator, including the demand response reward and fee profits Constraint (17) reflects the relationship between charging profit and charging load, constraint (18) is the method for calculating demand response rewards, and constraint (19) represents the range of charging prices set by electric vehicle fast charging station operators.

[0042] As a further improvement of the present invention, the demand response information in step 2 includes key information of the electric vehicle fast charging station intelligent body and key information of the distribution network intelligent body.

[0043] As a further improvement of the present invention, the key information of the electric vehicle fast charging station intelligent body is shown as follows;

[0044] 1. Status;

[0045] The state observed by the agent consists of charging price, charging load, distribution network electricity price at the last time section, and demand response information:

[0046]

[0047] 2. Action;

[0048] The action output is the charging price of the electric vehicle fast charging station. Specifically, the policy network will output the optimal value of the charging price to maximize its own profit:

[0049]

[0050] 3. Rewards;

[0051] The agent's reward is the profit of the EV fast charging station, which is determined by the charging load generated by the traffic allocation model taking into account user equilibrium. To maximize the reward, the policy network selects an appropriate charging price to increase the profit of the EV fast charging station.

[0052]

[0053] As a further improvement of the present invention, the key information of the distribution network intelligent entity is shown as follows;

[0054] 1) Status;

[0055] The distribution network status consists of the distribution network operation information: charging load, node voltage, and the amount of electricity purchased from the upstream grid at the current time;

[0056]

[0057] 2) Action;

[0058] The action of the distribution network agent is the demand response information;

[0059]

[0060] 3) Rewards;

[0061] The distribution network operator’s reward is the algebraic sum of the distribution network operating costs and the demand response costs paid to the EV fast charging stations;

[0062]

[0063] As a further improvement of the present invention, the step 4 of solving the traffic assignment model taking into account user equilibrium includes:

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072] In the formula, x a and x k represents the electric vehicle flow of road section a and electric vehicle charging station k, respectively, t a (x a ) and t k (x k ) represent the travel time of road section a and the waiting time of charging queue at electric vehicle charging station k, respectively, p,w represents the traffic flow on path p of the start-end point pair w, d w represents the electric vehicle travel demand for the start-end point pair w, and They represent the free travel time of road section a and the inherent charging service time of charging station k respectively; c a and c k The traffic capacity of road section a and the charging capacity of charging station k; and v k,p,w The coupling relationship between road section a, charging station k and path p is pointed out, and ω is the conversion coefficient of time and money; The charging load of charging k, Represents the amount of electricity needed to replenish the electric vehicle through path p;

[0073] Constraint (26) is the objective function of the traffic flow distribution model taking into account the user game equilibrium, including travel time cost, charging waiting cost and charging cost. Constraint (27) explains the coupling relationship between path traffic flow and section traffic flow. Constraint (28) clarifies the connection between path traffic flow and the number of vehicles waiting at electric vehicle fast charging stations. Constraint (29) follows the classic BRP equation and describes the interaction between link flow and link passing time. Constraint (30) outlines the calculation method for calculating the waiting time of electric vehicle fast charging stations. Constraint (31) summarizes the corresponding relationship between path traffic flow and electric vehicle travel demand. Constraint (32) ensures the non-negativity of path traffic flow. Constraint (33) reflects the relationship between the charging load of electric vehicle fast charging stations and path traffic flow.

[0074] As a further improvement of the present invention, in the process of training and decision-making of the traffic assignment model in step 4, in order to ensure that the safe operation constraints of the power grid are not violated, a CPO algorithm is used to obtain the optimal decision, including:

[0075] The optimal action strategy is obtained by the CPO algorithm. To illustrate the principle of the CPO algorithm, consider the following compact model, whose objective function is defined as follows;

[0076]

[0077]

[0078]

[0079] In the formula, π k is the current strategy, According to the strategy π k and The constraint cost when sampling states and actions, C is the average KL divergence, defined as Here, for sufficiently small By using first-order approximations for the objective and constraints, and second-order approximations for the trust region, (34)-(36) are written as;

[0080] θ k+1 = argmax θ y T (θ-θ k ) (37)

[0081] st c+b T (θ-θ k )≤0 (38)

[0082]

[0083] In the formula, Corresponding to the true reward function, Corresponding to the constraint cost function, H represents

[0084] In order to obtain the optimal action strategy in a given environment, the CPO algorithm operates by interacting with the reward and cost functions. The objective function of the CPO algorithm is defined as follows (40);

[0085] When the above models (37)-(39) are feasible, the update rule is derived by solving the following problem:

[0086] min θ c+b T (θ-θk )(40)

[0087]

[0088]

[0089] Let x = θ - θ k , then its dual function L[x,λ] is given by:

[0090]

[0091] The x obtained above should satisfy the trust domain constraint:

[0092]

[0093] Therefore, the parameter update rule when model (2-37)-(2-39) is infeasible is:

[0094]

[0095] When models (2-37)-(2-39) are feasible, the analytical solution to the above problem is as follows;

[0096]

[0097]

[0098]

[0099]

[0100] Compared with the prior art, the present invention has the following beneficial effects:

[0101] The present invention provides a method for mining the demand response potential of electric vehicle fast charging stations. Considering that private electric vehicle fast charging station operators are not sensitive to traffic network information (such as electric vehicle travel demand, etc.), in order to avoid possible loss of charging station benefits and waste of demand response potential, the interaction process between the charging station operator and the environment is modeled as a Markov chain, and the SAC (soft actor critic) algorithm is used to obtain its optimal strategy. At the same time, in order to avoid the demand response information output by the power grid causing the charging load to violate the power grid security constraints, the CPO (constrained policy optimization) algorithm is used to obtain the optimal strategy of the power grid. This method deeply mines the demand response potential of charging stations and is expected to provide new ideas for electric vehicles to participate in power grid scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] Figure 1 It is a patent flow chart;

[0103] Figure 2 It is a charging load comparison chart;

[0104] Figure 3 This is a comparison chart of charging station revenue. DETAILED DESCRIPTION

[0105] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments;

[0106] The present invention provides a process such as Figure 1 As shown, a method for mining the demand response potential of electric vehicle fast charging stations, the specific steps are as follows:

[0107] In the first step, the distribution network, electric vehicle fast charging stations, and electric vehicle users in the transportation network are regarded as different stakeholders, and since the distribution network and charging station operators are partially unknown to the external environment, their decision-making process is modeled as a Markov chain.

[0108] In the second step, the distribution network operator obtains the distribution network operation parameters by solving the distribution network Distflow model and broadcasts the formulated demand response information to all electric vehicle charging station operators.

[0109] Solve the distribution network Distflow model to obtain operating parameters and output demand response information, including:

[0110] Consider having The radial topology of the distribution network is represented by represents the set of lines in the distribution network g, and Represents the set of buses in the distribution network g. By solving the following linear Disflow model, the operating parameters of the distribution network can be obtained.

[0111]

[0112]

[0113]

[0114]

[0115]

[0116]

[0117]

[0118]

[0119] Where P g,ijand Q g,ij represents the active / reactive power flow of line (i, j) in the distribution network g; is an auxiliary variable. The representative node j is the root node of the distribution network g; r g,ij and x g,ij are the resistance and reactance of line (i, j) respectively; U g,j , and Indicates the voltage amplitude, and its upper and lower limits; and Represents the total power load, conventional power load, and charging load; and It is the active / reactive power output and power factor of photovoltaic power generation.

[0120] Constraints (1)-(2) ensure the balance of active and reactive power at each node in the distribution network. Constraint (3) represents the relationship between voltage drop and active / reactive power on the line. Constraints (4)-(5) consider the power factor of photovoltaic power generation. At the same time, equipment constraints are imposed on the active and reactive power output of the PV output. Constraint (6) represents the randomness of the PV active output and obeys a certain probability distribution. Constraint (7) summarizes the node power load, where the charging load It is determined by solving the traffic assignment model. Constraint (8) limits the range of voltage amplitude.

[0121] After obtaining the charging load of the EV fast charging station, the distribution network operator solves the linearized Distflow model to obtain detailed operation data. After obtaining the distribution network operation details, the charging load is adjusted through the generated demand response information, which is then sent to the EV fast charging station operator.

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129] Where g, k and i represent the indexes of distribution network, electric vehicle fast charging station and distribution network line respectively; G is the set of distribution networks, K is the set of electric vehicle fast charging stations; C is the total cost of the distribution network operator, including operating cost and demand response cost. represents the operating cost of the distribution network g, including the electricity charges for photovoltaic power generation cut-off and transformers, C DR represents the demand response cost for distribution network operators; represents the upper and lower demand response limits imposed by the distribution network operator on the billable load EV fast charging station k; represents the charging load k of the electric vehicle fast charging station; π e and π pv They represent the demand response electricity price, the electricity purchase price from the superior power grid and the penalty for underutilizing photovoltaic active power.

[0130] Constraint (9) is the objective function of the distribution network operator, including the distribution network operating cost and demand response fee. Constraints (10)-(12) are the methods for calculating the demand response cost and operating cost. Constraints (13)-(15) show the feasible area of ​​the demand response information formulated by the distribution network operator.

[0131] The operational constraints and profit objectives of charging station operators include:

[0132] We consider the optimal pricing strategy for EV charging stations. The electricity selling price of the distribution network operator is considered as the input of this process, while the charging price of the EV fast charging station is the output. The established charging price is then broadcasted to all EV users. The EV fast charging station uses environmental information, including historical charging load and electricity prices, to determine a reasonable charging price while maximizing revenue:

[0133]

[0134]

[0135]

[0136]

[0137] In the formula, H k represents the profit of electric vehicle fast charging station; k ,λ min ,λ max represents the charging price and its upper and lower limits; γ g,k is an auxiliary variable that specifies the connection relationship between the distribution network nodes and the electric vehicle fast charging stations.

[0138] Constraint (16) is the objective function of the EV fast charging station operator, including the demand response reward and fee profits Constraint (17) reflects the relationship between charging profit and charging load. Constraint (18) is the method for calculating demand response rewards. Constraint (19) represents the range of charging prices set by electric vehicle fast charging station operators.

[0139] The key information of the electric vehicle fast charging station intelligent entity is shown below.

[0140] 1. Status;

[0141] The state observed by the agent consists of charging price, charging load, distribution network electricity price at the last time section, and demand response information:

[0142]

[0143] 2. Action;

[0144] The action output is the charging price of the electric vehicle fast charging station. Specifically, the policy network will output the optimal value of the charging price to maximize its own profit:

[0145]

[0146] 3. Rewards;

[0147] The agent's reward is the profit of the EV fast charging station, which is determined by the charging load generated by the traffic allocation model taking into account user equilibrium. To maximize the reward, the policy network can choose an appropriate charging price to increase the profit of the EV fast charging station.

[0148]

[0149] The key information of the distribution network intelligent agent is shown below.

[0150] 1) Status;

[0151] The distribution network status consists of the distribution network operation information: charging load, node voltage, and the amount of electricity purchased from the upstream power grid at the current time.

[0152]

[0153] 2) Action;

[0154] The action of the distribution network intelligent agent is the demand response information.

[0155]

[0156] 3) Rewards;

[0157] The distribution network operator's reward is the algebraic sum of the distribution network operating costs and the demand response costs paid to EV fast charging stations.

[0158]

[0159] In the third step, after the charging station operator observes the historical and current distribution network electricity prices and demand response information, it outputs the charging price of each charging station.

[0160] The fourth step is to obtain the final result of the user game by solving the traffic distribution model that takes into account user equilibrium, and return the charging load to the distribution network and charging stations.

[0161] Solve traffic assignment models that take into account user equilibrium, including:

[0162]

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169]

[0170] In the formula, x a and x k They represent the electric vehicle flow of road section a and electric vehicle charging station k respectively. a (x a ) and t k (x k ) represent the travel time of road section a and the waiting time of charging queue at electric vehicle charging station k respectively. p,w represents the traffic flow on path p of the start-end point pair w. w represents the electric vehicle travel demand for the start-end point pair w. and They represent the free travel time of road section a and the inherent charging service time of charging station k respectively; c a and c k The traffic capacity of road section a and the charging capacity of charging station k; and v k,p,wThe coupling relationship between road section a and charging station k and path p is pointed out. ω is the conversion coefficient of time and money; The charging load of charging k, Represents the amount of electricity needed to replenish the electric vehicle through path p.

[0171] Constraint (26) is the objective function of the traffic flow allocation model taking into account the user game equilibrium, including travel time cost, charging waiting cost and charging cost. Constraint (27) illustrates the coupling relationship between path traffic flow and section traffic flow. Constraint (28) clarifies the connection between path traffic flow and the number of vehicles waiting at electric vehicle fast charging stations. Constraint (29) follows the classic BRP equation and describes the interaction between link flow and link passing time. Constraint (30) outlines the calculation method for calculating the waiting time of electric vehicle fast charging stations. Constraint (31) summarizes the corresponding relationship between path traffic flow and electric vehicle travel demand. Constraint (32) ensures the non-negativity of path traffic flow. Constraint (33) reflects the relationship between the charging load of electric vehicle fast charging stations and path traffic flow.

[0172] Using historical typical daily travel demand and charging prices, we adopt normal distribution for sampling to obtain a large number of sample features, solve the above model based on commercial solvers (such as GURUBI, CPLEX, etc.), obtain historical path traffic as labels, build a database, and use transformer with multi-head attention mechanism to minimize training error.

[0173] In order to ensure that the distribution network agent does not violate the safe operation constraints of the power grid during the training and decision-making process, the CPO algorithm is used to obtain the optimal decision, including:

[0174] This patent uses the CPO algorithm to obtain the optimal action strategy. To illustrate the principle of the CPO algorithm, consider the following compact model, whose objective function is defined as follows.

[0175]

[0176]

[0177]

[0178] In the formula, π k is the current strategy, According to the strategy π k and The constraint cost when sampling states and actions, C is the average KL divergence, defined as Here, for sufficiently small By using first-order approximations for the objective and constraints, and second-order approximations for the trust region, (34)-(36) can be written as

[0179] θ k+1 = argmax θ y T (θ-θ k ) (37)

[0180] st c+b T (θ-θ k )≤0 (38)

[0181]

[0182] In the formula, Corresponding to the true reward function, Corresponding to the constraint cost function, H represents D KL .

[0183] In order to obtain the optimal action strategy in a given environment, the CPO algorithm operates by interacting with reward and cost functions. The objective function of the CPO algorithm is defined as follows (40).

[0184] When the above models (37)-(39) are feasible, the update rule is derived by solving the following problem:

[0185] min θ c+b T (θ-θ k )(40)

[0186]

[0187]

[0188] Let x = θ - θ k , then its dual function L[x,λ] is given by:

[0189]

[0190] The x obtained above should satisfy the trust domain constraint:

[0191]

[0192] Therefore, the parameter update rule when model (2-37)-(2-39) is infeasible is:

[0193]

[0194] When models (2-37)-(2-39) are feasible, the analytical solution to the above problem is as follows.

[0195]

[0196]

[0197]

[0198]

[0199] Step 5: Iterate steps 2 to 4 repeatedly until the reward curve converges.

[0200] Results

[0201] 1. Comparison of charging load control results

[0202] exist Figure 2 In this paper, we compare three different charging load regulation methods. Between 11 am and 2 pm, distribution network operators try to adapt to renewable energy by increasing charging load. To achieve this goal, the marginal price-based method sets the charging price at a relatively low level to attract more electric vehicle users to charge. When the grid is heavily burdened, especially between 5 pm and 8 pm, the marginal price-based method will significantly increase the charging price. It is not difficult to see that although this strategy can effectively reduce the peak charging load, it does not fully consider the profitability of charging stations.

[0203] On the other hand, in the absence of a demand response mechanism, charging station operators are mainly concerned with maximizing their own economic benefits, especially during periods of high electric vehicle traffic, when the charging load is significantly higher. The demand response mechanism proposed in this patent successfully combines the advantages of the first two methods. This method sets a higher demand response cap when the charging load needs to be increased, encouraging charging stations to set low prices to attract more users to charge. When the coupled network load is heavy, a lower demand response cap will be set to reduce overload during peak hours.

[0204] 2. Comparison of Charging Station Revenue

[0205] In order to motivate charging station operators to actively participate in the implementation of distribution network demand response policies, it is necessary to ensure that the economic benefits of charging stations are no less than the profits obtained through non-demand response methods, such as Figure 3 Therefore, if the distribution network implements a demand response strategy to manage the charging load, appropriate economic compensation must be provided to the charging station operators to offset the economic losses caused by the change in charging load.

[0206] The above description is only a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent change made based on the technical essence of the present invention still falls within the scope of protection required by the present invention.

Claims

1. A method for mining the demand response potential of electric vehicle fast charging stations, characterized in that: The specific steps are as follows: In the first step, the distribution network, electric vehicle fast charging stations, and electric vehicle users in the transportation network are regarded as different stakeholders. Since the distribution network and charging station operators are partially unknown to the external environment, their decision-making process is modeled as a Markov chain. In the second step, the distribution network operator obtains the distribution network operation parameters by solving the distribution network Distflow model and broadcasts the formulated demand response information to all electric vehicle charging station operators; Solve the distribution network Distflow model to obtain operating parameters and output demand response information, including: Consider having The radial topology of the distribution network is represented by represents the set of lines in the distribution network g, and Represents the bus set in the distribution network g. The operating parameters of the distribution network are obtained by solving the following linear Disflow model; Where P g,ij and Q g,ij represents the active / reactive power flow of line (i, j) in the distribution network g; and Represents the injected active and reactive power of the upper power grid; P g,is and Q g,is Represents the active and reactive power flows of the line; is an auxiliary variable. The representative node j is the root node of the distribution network g; r g,ij and x g,ij are the resistance and reactance of line (i, j) respectively; U g,j , and Indicates the voltage amplitude, and its upper and lower limits; and Represents the total power load, conventional power load, and charging load; and Active / reactive power output and power factor of photovoltaic power generation; Constraints (1)-(2) ensure the balance of active power and reactive power at each node in the distribution network, constraint (3) represents the relationship between voltage drop and active / reactive power on the line, and constraints (4)-(5) consider the power factor of photovoltaic power generation. At the same time, equipment constraints are imposed on the active and reactive power output of photovoltaic power. Constraint (6) represents the randomness of photovoltaic active output and obeys a certain probability distribution. Constraint (7) summarizes the node power load, where the charging load It is determined by solving the traffic assignment model. Constraint (8) limits the range of voltage amplitude; After obtaining the charging load of the EV fast charging station, the distribution network operator solves the linearized Distflow model to obtain detailed operation data. After obtaining the distribution network operation details, the charging load is adjusted through the generated demand response information, which is then sent to the EV fast charging station operator; Where g, k and i represent the indexes of distribution network, electric vehicle fast charging station and distribution network line respectively; G is the set of distribution network, K is the set of electric vehicle fast charging stations; C is the total cost of distribution network operator, including operating cost and demand response cost. represents the operating cost of the distribution network g, including the electricity charges for photovoltaic power generation cut-off and transformers, C DR represents the demand response cost for distribution network operators; represents the upper and lower demand response limits imposed by the distribution network operator on the billable load; Indicates the charging load of the electric vehicle fast charging station; π e and π pv They represent the demand response price, the price of electricity purchased from the upper grid, and the penalty for underutilizing the active power of photovoltaic power; Constraint (9) is the objective function of the distribution network operator, including the distribution network operating cost and demand response fee, constraints (10)-(12) are the methods for calculating the demand response cost and operating cost, and constraints (13)-(15) show the feasible area of ​​the demand response information formulated by the distribution network operator; In the third step, after observing the historical and current distribution network electricity prices and demand response information, the charging station operator outputs the charging price of each charging station; The fourth step is to obtain the final result of the user game by solving the traffic distribution model that takes into account user equilibrium, and return the charging load to the distribution network and charging stations; Step 5: Iterate steps 2 to 4 repeatedly until the reward curve converges.

2. The method for mining the demand response potential of an electric vehicle fast charging station according to claim 1, characterized in that: The operation constraints and profit objectives of the charging station operator in the objective function of the distribution network operator in step 2 include: Considering the optimal pricing strategy for electric vehicle charging stations, the electricity selling price of the distribution network operator is regarded as the input of this process, and the charging price of the electric vehicle fast charging station is regarded as the output. Subsequently, the charging price is broadcast to all electric vehicle users. The electric vehicle fast charging station will use environmental information, including historical charging load and electricity price, to determine a reasonable charging price, while maximizing the revenue: In the formula, H k represents the profit of electric vehicle fast charging station; k ,λ min ,λ max represents the charging price and its upper and lower limits; γ g,k is an auxiliary variable that specifies the connection relationship between the distribution network nodes and the EV fast charging stations; Constraint (16) is the objective function of the EV fast charging station operator, including the demand response reward and fee profits Constraint (17) reflects the relationship between charging profit and charging load, constraint (18) is the method for calculating demand response rewards, and constraint (19) represents the range of charging prices set by electric vehicle fast charging station operators.

3. The method for mining the demand response potential of an electric vehicle fast charging station according to claim 1, characterized in that: The demand response information in step 2 includes key information of the electric vehicle fast charging station intelligent entity and key information of the distribution network intelligent entity.

4. The method for mining the demand response potential of an electric vehicle fast charging station according to claim 3, characterized in that: The key information of the electric vehicle fast charging station intelligent agent is shown as follows; 1. Status; The state observed by the agent consists of charging price, charging load, distribution network electricity price at the last time section, and demand response information:

2. Action; The action output is the charging price of the electric vehicle fast charging station. Specifically, the policy network will output the optimal value of the charging price to maximize its own profit:

3. Rewards; The agent's reward is the profit of the EV fast charging station, which is determined by the charging load generated by the traffic allocation model taking into account user equilibrium. To maximize the reward, the policy network selects an appropriate charging price to increase the profit of the EV fast charging station. λ k,t-1 represents the charging price of charging station k at time t-1; Represents the electricity sales price of the distribution network operator; Represents the demand response electricity price of the distribution network operator during period t.

5. The method for mining the demand response potential of an electric vehicle fast charging station according to claim 3, characterized in that: The key information of the distribution network intelligent entity is shown as follows: 1) Status; Distribution network status is determined by distribution network operation information Composition: charging load, node voltage, and the amount of electricity purchased from the upstream power grid at the current time; 2) Action; The action of the distribution network agent is the demand response information; 3) Rewards; The distribution network operator’s reward is the algebraic sum of the distribution network operating costs and the demand response costs paid to the EV fast charging stations; 6. The method for mining the demand response potential of an electric vehicle fast charging station according to claim 1, characterized in that: The step 4 of solving the traffic assignment model taking into account user equilibrium includes: In the formula, x a and x k represents the electric vehicle flow of road section a and electric vehicle charging station k, respectively, t a (x a ) and t k (x k ) represent the travel time of road section a and the waiting time of charging queue at electric vehicle charging station k, respectively, p,w represents the traffic flow on path p of the start-end point pair w, d w represents the electric vehicle travel demand for the start-end point pair w, and They represent the free travel time of road section a and the inherent charging service time of charging station k respectively; c a and c k The traffic capacity of road section a and the charging capacity of charging station k; and v k,p,w The coupling relationship between road section a, charging station k and path p is pointed out, and ω is the conversion coefficient of time and money; The charging load of charging k, Represents the amount of electricity needed to replenish the electric vehicle through path p; Constraint (26) is the objective function of the traffic flow distribution model taking into account the user game equilibrium, including travel time cost, charging waiting cost and charging cost. Constraint (27) explains the coupling relationship between path traffic flow and section traffic flow. Constraint (28) clarifies the connection between path traffic flow and the number of vehicles waiting at electric vehicle fast charging stations. Constraint (29) follows the classic BRP equation and describes the interaction between link flow and link passing time. Constraint (30) outlines the calculation method for calculating the waiting time of electric vehicle fast charging stations. Constraint (31) summarizes the corresponding relationship between path traffic flow and electric vehicle travel demand. Constraint (32) ensures the non-negativity of path traffic flow. Constraint (33) reflects the relationship between the charging load of electric vehicle fast charging stations and path traffic flow.

Images