Charging strategy optimization method and system considering competition relationship of charging stations

Abstract

The invention discloses a charging strategy optimization method considering the competitive relationship of charging stations, and the method comprises the following steps: constructing a finite rationality value model considering the risk attitude and price sensitivity of a user, and depicting the selection preference of an electric vehicle user among different charging stations; establishing a master-slave game model for a plurality of charging stations with a competitive relationship; iteratively solving each sub-problem in the master-slave game model by adopting an adaptive alternating direction multiplier (ADMM) method to obtain the optimal time-of-use electricity price of the charging station; based on the optimal time-of-use electricity price of a charging station, an electric vehicle charging elastic clustering optimization model is established in a sliding time window, a large number of vehicles in different states are subjected to aggregation optimization according to the remaining charging duration, and finally an optimal charging strategy is obtained. Compared with a traditional strategy based on a fixed electricity price, the strategy optimization of price signal-user selection-charging power-load curve is realized in a competitive environment, and the utilization rate of renewable energy sources and the operation income of the charging station are improved.

Description

Technical Field

[0001] This invention relates to a charging strategy optimization method and system that considers the competitive relationship of charging stations. It relates to the fields of power system dispatching and electric vehicle charging management, and particularly to an electric vehicle charging strategy optimization method and system that considers market competition and uncertainty in the context of the integration of smart grids and transportation networks. Background Technology

[0002] With the widespread adoption of electric vehicles (EVs), their disorderly and concentrated charging behavior poses a serious challenge to the stable operation of the power grid. To guide EVs towards orderly charging and alleviate peak-valley pressure on the power grid, charging strategy optimization technology has become a current research hotspot. Traditional charging strategy optimization methods are mostly based on fixed or pre-set electricity price mechanisms. In these strategies, the power grid or charging station operators formulate a single, fixed electricity price signal based on predicted load conditions, and then use this price signal to centrally or in a distributed manner optimize and schedule the charging behavior of EV users. These methods can, to some extent, shift charging load and smooth out peaks and valleys. However, their core flaw lies in neglecting a crucial dynamic factor in the charging service market: the competitive relationship between charging stations.

[0003] However, formulating and implementing optimal charging strategies in a competitive market faces several challenges. First, the operating environment is significantly uncertain. Charging station operations are affected by various uncertainties, including the volatile output of distributed renewable energy sources and the stochastic behavior of electric vehicle users. Second, user behavior is complex. In reality, users consider not only price when choosing a charging station but also factors such as driving distance and expected waiting time, making the demand-price relationship highly nonlinear and nonconvex. Finally, the problem-solving complexity of multi-station competition and strategy optimization is high. Pricing games among multiple charging stations constitute a complex multi-agent decision-making problem. Traditional optimization algorithms are often inefficient in handling such problems, prone to getting trapped in local optima, and may even fail to derive feasible strategies. Summary of the Invention

[0004] The purpose of this invention is to address the challenges in optimizing electric vehicle charging strategies that consider the competitive relationship between charging stations, such as high uncertainty, bounded rationality of user decisions, and complexity of solving multi-station game problems. This invention proposes a charging strategy optimization method and system that considers the competitive relationship between charging stations. In a competitive environment, the optimal day-ahead time-of-use price is obtained through the Adaptive Alternating Direction Multiplier Method (ADMM). Based on this, an electric vehicle charging elastic clustering optimization model and reinforcement learning algorithm are introduced to refine the charging strategy within the stations, thereby achieving synergistic optimization of price and charging strategy.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A charging strategy optimization method considering the competition among charging stations includes the following steps:

[0007] Construct a bounded rationality value model that considers users' risk attitudes and price sensitivity to characterize electric vehicle users' choice preferences among different charging stations;

[0008] For multiple competing charging stations, a master-slave game model is established;

[0009] The adaptive alternating direction multiplier method (ADMM) is used to iteratively solve the subproblems in the master-slave game model to obtain the optimal time-of-use electricity price for the charging station.

[0010] Based on the optimal time-of-use electricity price of charging stations, an elastic clustering optimization model for electric vehicle charging is established within a sliding time window. This model aggregates and optimizes a large number of vehicles in different states according to their remaining charging time, ultimately obtaining the optimal charging strategy.

[0011] Furthermore, the above method specifically includes the following steps:

[0012] Step 1: Obtain the total number of electric vehicles, establish a charging station selection model for electric vehicle users based on bounded rationality, and determine the charging station demand function under different pricing strategies.

[0013] Step 2: Obtain the total number of competing charging stations and establish a master-slave game model with the target charging station as the leader and other charging stations as followers;

[0014] Step 3: Using the sampling average approximation SAA method, the uncertainties of random arrival of electric vehicle users and output of distributed renewable energy are discretized, and a set of finite representative operating scenarios are generated through simulation.

[0015] Step 4: To solve the master-slave game model established in Step 2, the Alternating Direction Multiplier Method (ADMM) framework is introduced. Through global consensus variables and consistency constraints, the original problem is decomposed into multiple parallel local optimization subproblems of charging stations.

[0016] Step 5: Based on the augmented Lagrangian function, alternately and iteratively update the local pricing strategy, global consensus price, and dual variables of each charging station until the algorithm converges and the optimal intraday time-of-use charging price is obtained.

[0017] Step 6: Establish a charging strategy optimization model based on electric vehicle charging elastic clustering. The daily time-of-use charging price, remaining power of charging stations, distributed wind and solar power generation, and the power required by electric vehicles obtained in Step 5 are used as inputs. Elastic clustering and aggregation modeling are performed on electric vehicles under different states within a sliding time window to obtain the optimal charging strategy for each time period.

[0018] Furthermore, the electric vehicle user charging station selection model established in step S1 takes charging price, driving distance, and expected queuing time as inputs.

[0019] Furthermore, the specific method for establishing the electric vehicle user charging station selection model based on bounded rationality in step S1 includes:

[0020] Set the total number of electric vehicles to ;

[0021] S1-1, Electric Vehicle Users At the charging station The overall value is expressed as:

[0022]

[0023] in, , and For attribute preference weights, This represents the relative importance that each user assigns to the charging price. This represents the relative importance that each user assigns to travel distance. This represents the relative importance that each user assigns to the waiting time, and ;

[0024] For price deviation, , It is the user's expected price. It is the first The charging price offered by each charging station;

[0025] Due to travel time deviation, , This indicates the user's expected travel time. Indicates if the first Electric vehicle selection The actual driving time for charging at a charging station;

[0026] Due to queuing time deviation, , This indicates the user's expected waiting time. Indicates if the first Electric vehicle selection The actual waiting time for charging at each charging station; Represents the utility function;

[0027] S1-2, using the S-shape function Objective deviation Mapped to subjective psychological values, the following equation is established:

[0028]

[0029] in, Indicates attribute deviation; It is a function of decreasing return sensitivity. This is the loss sensitivity reduction coefficient. , ; Indicates the loss aversion coefficient. ;

[0030] S1-3, waiting time is estimated using the M / M / c queuing model:

[0031]

[0032]

[0033] in, Indicates the number of charging stations; This represents the average arrival rate of EV users; This represents the average service rate of a single charging station. Indicates charging station utilization rate ; Indicates charging station The probability that all charging stations within the area are idle; This indicates the number of charging piles occupied within the charging station;

[0034] S1-4, simulating users through polynomial Logit. The user's charging station selection behavior Choose a charging station The probability is:

[0035]

[0036] in, This represents a collection of all charging stations. Indicates user Choose a charging station The effect, express The index value;

[0037] Based on the probability distribution, the user's final choice of charging station is determined, then the... Electric vehicles at charging station The required charging energy is calculated as follows:

[0038]

[0039] in, Indicates the first The current state of charge (SOC) of the electric vehicle at time t; Indicates the first The battery capacity of an electric vehicle; This indicates the electricity consumption of an electric vehicle per kilometer. This indicates the distance between the electric vehicle and the charging station.

[0040] Furthermore, the specific implementation method of step 2 includes:

[0041] S2-1, Determine the set of competing charging stations Upper price limit Price lower bound Set pricing time With scene set ,in This represents the total number of time periods;

[0042] S2-2, Target charging station As the leader, the rest of the sites As a follower, the leader first announces the price vector. Each follower observed Then each person selects their own price. Subsequently, the user charging station selection model provides information for each station in uncertain scenarios. The following needs ,in Indicates the leader exist The price of a moment Indicates that the followers are The price of a moment;

[0043] S2-3, for any follower Establishing a profit maximization problem:

[0044]

[0045]

[0046]

[0047] in, This indicates the operating costs of competing charging stations; This indicates the amount of electricity purchased from the power grid; This indicates the unit cost of purchasing electricity. This indicates the potential losses from wind power generation. This indicates the potential losses from solar power generation; and This refers to distributed wind power output and distributed photovoltaic power output; Indicates the charging power of electric vehicles; Indicates in Always need to be at a charging station The number of electric vehicles being charged;

[0048] S2-4, the leader solves for profit maximization under the condition of anticipating the joint response of followers:

[0049]

[0050]

[0051] in, Indicates an uncertain scenario Next One charging station The constant demand for electric vehicle charging.

[0052] Furthermore, the specific implementation method of step 3 includes:

[0053] The Sample Average Approximation (SAA) method is used to handle uncertainties in wind and solar power generation. The local objective function of a charging station is expressed as:

[0054]

[0055]

[0056] in, Representing a scene Next The profit of each charging station.

[0057] Furthermore, the specific implementation method of step 4 includes:

[0058] S4-1 decomposes the multi-station game into There are several parallel solvable subproblems, each corresponding to a separate charging station;

[0059] S4-2, Alternating Direction Multiplier Method (ADMM) introduces auxiliary variables. ( The core mechanism for achieving decomposition is to ensure consistency constraints across charging stations. , Indicates charging station Local decision variables.

[0060] Furthermore, the specific implementation method of step 5 includes:

[0061] S5-1, Based on the objective function of formula (13) and the consistency constraints described in step 4, the charging station The augmented Lagrangian function is:

[0062]

[0063] in, To constrain Related dual variables, For penalty parameters;

[0064] S5-2, local price optimization, parallel computing update :

[0065]

[0066] This yields the optimal price vector for each charging station in this iteration;

[0067] S5-3, Calculate the new consensus price and update the dual variables based on the local update results from each site:

[0068]

[0069]

[0070] S5-4, Calculate the original residuals and dual residuals:

[0071]

[0072]

[0073] In S5-5, to accelerate convergence, an adaptive penalty parameter is introduced, which is dynamically adjusted based on the magnitudes of the original residual and the dual residual. :

[0074]

[0075] in, , and Indicates the parameter, set to , ;

[0076] The optimal charging price for the charging station is obtained by continuing to update and iterate according to formula (15) until convergence.

[0077] Furthermore, the specific implementation method of step 6 includes:

[0078] S6-1, Based on the optimal price in step 5 and the amount of electricity required for charging the electric vehicle in step 1. Based on the charging strategy at the previous moment, calculate the remaining charging amount needed for each electric vehicle at the charging station at the current moment. And set the sliding time. ;

[0079] S6-2, At the current moment, based on the charging station status, compare the charging time required and the remaining time for all electric vehicles in the charging station, and then aggregate and classify the electric vehicles at the current moment:

[0080]

[0081]

[0082] in, Indicates the first The rated charging power of each charging station Indicates aggregation and classification. Indicates the first The remaining charging time required for a vehicle. Indicates rounding up;

[0083] S6-3, for electric vehicles within each cluster, establish the following equation to determine the surplus time of electric vehicles within the cluster. Electric vehicles with less surplus time have higher charging priority:

[0084]

[0085] in, Indicates the first The surplus time of each electric vehicle in the aggregate;

[0086] S6-4, Set action sequences for aggregated classification. The action sequence represents the overall charging probability of electric vehicles in each aggregate, where , ;

[0087] Based on the current state Let be the proportion of charging piles and the proportion of wind and solar power generation in the charging station, respectively. Then, we establish the wind and solar power ratio and the profit maximization problem (i.e., the objective function). The established objective function is as follows:

[0088]

[0089]

[0090] in, Represents the weighting coefficient, which is The number between; Indicates action strategy; This represents the number of electric vehicles in the k-th aggregate at the current time;

[0091] S6-5, using Q-learning in The following Q-value iterative formula is established: based on the current state, the behavior with the maximum Q-value is obtained through iterative learning as the optimal charging behavior.

[0092]

[0093] in, For learning rate, As a discount factor, This indicates the total number of time periods.

[0094] S6-6, determine the priority of electric vehicles according to formula (23), determine the number of electric vehicles selected for charging in each cluster according to the optimal charging behavior and priority of each cluster, and then obtain the optimal charging strategy by multiplying the number of charging vehicles by the charging power.

[0095] The present invention also provides a charging strategy optimization system that considers the competitive relationship between charging stations, the system comprising:

[0096] The User Choice and Demand Response module is used to construct a bounded rationality value model that considers users’ risk attitudes and price sensitivity, and to characterize the choice preferences of electric vehicle users among different charging stations.

[0097] The competitive pricing game construction module is used to establish a master-slave game pricing model for multiple competing charging stations.

[0098] The adaptive ADMM iterative solution module uses the adaptive alternating direction multiplier method (ADMM) to iteratively solve the sub-problems in the master-slave game model and obtain the optimal time-of-use electricity price for the charging station.

[0099] The charging elastic clustering and charging strategy solution module, based on the optimal time-of-use electricity price of the charging station obtained by the optimal time-of-use electricity price acquisition module, establishes an electric vehicle charging elastic clustering optimization model within a sliding time window;

[0100] The optimal charging strategy output module aggregates and optimizes a large number of vehicles in different states based on their remaining charging time. Specifically, it determines the number of electric vehicles selected for charging in each cluster based on the optimal charging behavior and priority of each cluster, and then obtains the optimal charging strategy by multiplying the number of vehicles by the charging power.

[0101] Furthermore, the above system is used to implement a charging strategy optimization method that takes into account the competitive relationship between charging stations.

[0102] The beneficial effects of this invention include:

[0103] Establish a user choice model based on bounded rationality to accurately characterize the irrational decision-making and price sensitivity of electric vehicle users, making demand forecasts more realistic.

[0104] The adaptive ADMM algorithm can handle the uncertainties of renewable energy output and user arrival, and improves numerical stability and convergence speed.

[0105] By constructing an elastic clustering and aggregation model for electric vehicle charging within a sliding time window, the charging probability distribution of large-scale electric vehicles under different states is characterized. Furthermore, reinforcement learning algorithms are used to obtain the optimal charging action strategy, thereby achieving fine-grained scheduling of charging power and charging time for various types of vehicles. Attached Figure Description

[0106] The present invention will be further described below with reference to the accompanying drawings:

[0107] Figure 1 This is a block diagram illustrating the principle of the electric vehicle charging strategy method of the present invention;

[0108] Figure 2 This is a diagram illustrating the intraday pricing after the process of bounded rationality and game theory pricing.

[0109] Figure 3 It is a graph showing the charging power and wind and solar power output at each moment in the charging station;

[0110] Figure 4 This is a schematic diagram of the charging strategy optimization system that considers the competition among charging stations in the embodiment. Detailed Implementation

[0111] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0112] Example 1:

[0113] See attached document Figure 1 This embodiment provides a charging strategy optimization method that considers the competition among charging stations. The method includes the following steps:

[0114] S1, Initialization, input competitive pricing parameters;

[0115] Input competitive charging station set Price upper and lower limits , Site parameters (number of charging piles) Service rate Distributed renewable energy data; target charging stations As the leader, the rest of the sites For followers.

[0116] S2, User Choice and Demand Response Construction;

[0117] We construct a bounded rationality value model that considers users’ risk attitude and price sensitivity, and use multinomial logit discrete choice to characterize electric vehicle users’ choice preferences among different charging stations.

[0118] Set the total number of electric vehicles to For any electric vehicle user Select site The overall value is expressed as:

[0119]

[0120] in, , and For attribute preference weights, This represents the relative importance that each user assigns to the charging price. This represents the relative importance that each user assigns to travel distance. This represents the relative importance that each user assigns to the waiting time, and ;

[0121] For price deviation, , It is the user's expected price. It is the first The charging price offered by each charging station;

[0122] Due to travel time deviation, , This indicates the user's expected travel time. Indicates if the first Electric vehicle selection The actual driving time for charging at a charging station;

[0123] Due to queuing time deviation, , This indicates the user's expected waiting time. Indicates if the first Electric vehicle selection The actual waiting time for charging at each charging station; Represents the utility function.

[0124] Using S-shaped value function Objective deviation Mapped to subjective psychological values, the following equation is established:

[0125]

[0126] in, Indicates attribute deviation; It is a function of decreasing return sensitivity. This is the loss sensitivity reduction coefficient. , ; Indicates the loss aversion coefficient. .

[0127] Waiting time is estimated using the M / M / c queuing model:

[0128]

[0129]

[0130] in, Indicates the number of charging stations; This represents the average arrival rate of EV users; This represents the average service rate of a single charging station. Indicates charging station utilization rate ; Indicates charging station The probability that all charging stations within the area are idle; This indicates the number of charging piles occupied within the charging station.

[0131] Simulate users using polynomial logit. The user's charging station selection behavior Choose a charging station The probability is:

[0132]

[0133] in, This represents a collection of all charging stations. Indicates user Choose a charging station The effect, express The index value;

[0134] Based on the probability distribution, the user's final choice of charging station is determined, then the... Electric vehicles at charging station The required charging energy is calculated as follows:

[0135]

[0136] in, Indicates the first The current state of charge (SOC) of the electric vehicle at time t; Indicates the first The battery capacity of an electric vehicle; This indicates the electricity consumption of an electric vehicle per kilometer. This indicates the distance between the electric vehicle and the charging station.

[0137] S3, competitive pricing game construction;

[0138] The leader announces the price vector first. Each follower observed Then each person selects their own price. Subsequently, the user charging station selection model provides information for each station in uncertain scenarios. The following needs ,in Indicates the leader exist The price of a moment Indicates that the followers are The price of a moment.

[0139] For any follower Establishing a profit maximization problem:

[0140]

[0141]

[0142] in, This indicates the operating costs of competing charging stations; This indicates the amount of electricity purchased from the power grid. ; This indicates the unit cost of purchasing electricity. This indicates the potential losses from wind power generation. This indicates the potential losses from solar power generation; and This refers to distributed wind power output and distributed photovoltaic power output; Indicates the charging power of electric vehicles; Indicates in Always need to be at a charging station The number of electric vehicles being charged.

[0143] Leaders solve for profit maximization by anticipating the joint response of followers:

[0144] (9)

[0145] (10)

[0146] in, Indicates an uncertain scenario Next One charging station The constant demand for electric vehicle charging.

[0147] S4, Constraint Construction;

[0148] Determine the upper and lower bounds of the price: Queue stability: ;

[0149] Energy balance: ;

[0150] in, Indicates wind power generation. Indicates the amount of solar power generated. This indicates the amount of electricity purchased from the power grid. Indicates time At the charging station The number of electric vehicles that need charging. This indicates the charging power of an electric vehicle.

[0151] S5, optimize the construction of the objective function;

[0152] Consider charging stations Taking profit maximization as the pricing optimization objective for charging stations, we have the following equation:

[0153] (11)

[0154] S6, scenario-based simulation;

[0155] The Sample Average Approximation (SAA) method is used to handle uncertainties in wind and solar power generation. The local objective function of a charging station can be expressed as:

[0156] (12)

[0157] (13)

[0158] in, Representing a scene Next The profit of each charging station.

[0159] S7, Adaptive ADMM iterative solution;

[0160] Adaptive ADMM introduces auxiliary variables ( The core mechanism for achieving decomposition is to ensure consistency constraints across charging stations. , Indicates charging station Local decision variables. Based on the above objectives and consistency constraints, the charging station... The augmented Lagrangian function is:

[0161] (14)

[0162] In the In this iteration, local price optimization is performed, and updates are performed using parallel computation. :

[0163] (15)

[0164] This yields the local optimal price vector for each charging station in the current iteration. Based on the local update results for each station, the new consensus price is calculated and the dual variable is updated:

[0165] (16)

[0166] (17)

[0167] Calculate the original residuals and dual residuals:

[0168] (18)

[0169] (19)

[0170] To accelerate convergence, an adaptive penalty parameter is introduced, which is dynamically adjusted based on the magnitudes of the original residual and the dual residual. :

[0171] (20)

[0172] in, , and Indicates the parameter, set to , By iterating repeatedly using formulas (15)-(20), until... , This allows us to obtain the optimal price vector for each charging station.

[0173] S8, charging elastic clustering and charging strategy solution;

[0174] Based on the optimal price in step S7 And the amount of electricity required for charging the electric vehicle in step S2. Based on the charging strategy from the previous moment, the remaining charge amount for each electric vehicle in the charging station at the current moment is calculated. And set the sliding time. ;

[0175] At the current moment, based on the charging station's status, the charging time required for all electric vehicles within the station is compared with the remaining time. If the required charging time for an electric vehicle is greater than the remaining time, it must be charged. If the required charging time is less than the remaining time, the required charging time is used to group the electric vehicles. The electric vehicles at the current moment are then aggregated and categorized.

[0176] (twenty one)

[0177] (twenty two)

[0178] in, Indicates the first The rated charging power of each charging station Indicates aggregation and classification. Indicates the first The remaining charging time required for a vehicle. This indicates rounding up to the nearest integer.

[0179] For each electric vehicle within an aggregate, the following equation is established to determine the surplus time of electric vehicles within the aggregate. Electric vehicles with shorter surplus time have higher charging priority:

[0180] (twenty three)

[0181] Set action sequences for aggregation categories. This represents the overall charging probability of electric vehicles in each aggregate, where , Then, based on the charging probability and the priority of electric vehicles within each aggregation, selection is made from highest to lowest priority. This is based on the current state: The problem involves determining the occupancy ratio of charging piles and wind / solar power generation in charging stations, and then establishing the wind / solar power ratio and maximizing profits.

[0182] (twenty four)

[0183] (25)

[0184] in, Represents the weighting coefficient, which is The number between Indicates action strategy, This represents the number of electric vehicles in the k-th aggregate at the current time. The objective function considers maximizing the profit of the charging station and maximizing the proportion of wind and solar power generation in the charging station.

[0185] Using Q-learning in The following Q-value iterative formula is established to obtain the optimal charging behavior:

[0186] (26)

[0187] in, For learning rate, This is the discount factor.

[0188] S9. Based on the optimal charging behavior and priority of each cluster, determine the number of electric vehicles selected for charging in that cluster, and then obtain the optimal charging strategy by multiplying the number of vehicles by the charging power.

[0189] Example 2:

[0190] This embodiment provides a charging strategy optimization system that considers the competition among charging stations. The system includes:

[0191] The User Choice and Demand Response module is used to construct a bounded rationality value model that considers users’ risk attitudes and price sensitivity, and to characterize the choice preferences of electric vehicle users among different charging stations.

[0192] The competitive pricing game construction module is used to establish a master-slave game pricing model for multiple competing charging stations.

[0193] The adaptive ADMM iterative solution module uses the adaptive alternating direction multiplier method (ADMM) to iteratively solve the sub-problems in the master-slave game model and obtain the optimal time-of-use electricity price for the charging station.

[0194] The charging elastic clustering and charging strategy solution module, based on the optimal time-of-use electricity price of the charging station obtained by the optimal time-of-use electricity price acquisition module, establishes an electric vehicle charging elastic clustering optimization model within a sliding time window;

[0195] The optimal charging strategy output module aggregates and optimizes a large number of vehicles in different states based on their remaining charging time. Specifically, it determines the number of electric vehicles selected for charging in each cluster based on the optimal charging behavior and priority of each cluster, and then obtains the optimal charging strategy by multiplying the number of vehicles by the charging power.

[0196] Furthermore, the system also includes an initialization module for inputting competitive pricing parameters.

[0197] Furthermore, the system also includes:

[0198] The constraint construction module is used to construct price upper bound, price lower bound, queue stability constraint and energy balance constraint based on operational requirements such as the number of charging piles, upper limit of charging power, wind and solar power output, grid power purchase capacity and queue stability, thereby limiting the feasible solution space for the charging station pricing game model.

[0199] The objective function construction module is used to set the objective function for charging station pricing optimization based on the constraints. Specifically, it takes the maximization of charging station operating profit as the time-of-use pricing optimization objective by comprehensively considering electricity purchase cost, renewable energy generation loss and charging service revenue.

[0200] The scenario-based simulation module is used to divide and simulate the random arrival behavior of electric vehicle users and the uncertainty of distributed wind power and photovoltaic power generation using the sample average approximation method, generating a set of representative operating scenarios to provide data support for subsequent competitive game strategy evaluation.

[0201] The competitive game strategy screening module is used to compare and screen different competitive pricing strategies based on the demand response of each candidate time-of-use electricity pricing strategy obtained from scenario simulation within a master-slave game framework, and output the competitive pricing strategy of the charging station that simultaneously satisfies the constraints and optimizes the objective function.

[0202] Example 3:

[0203] This embodiment provides an experiment conducted on the method of the present invention.

[0204] The test scenario included 1,000 electric vehicles in a traffic network, spanning 24 hours, and optimized the charging strategies of three competing charging stations.

[0205] Experimental results are as follows Figure 2 and Figure 3 As shown, where, Figure 2 The display shows the day-of-day time-of-use electricity prices for the three charging stations. Figure 3 It displays the charging power and wind / solar output curves at each moment in the charging station. From Figure 2 It can be seen that, after adaptive ADMM optimization, under conditions of uncertain arrival and renewable output, price-selection-load co-optimization can be achieved through competitive time-of-use pricing. Figure 3 It can be seen that by flexibly aggregating electric vehicle charging, wind and solar power generation can be better utilized, the waste of renewable energy can be reduced, and the operating profit of charging stations can be improved.

[0206] Simulation results show that the charging strategy optimization method proposed in this invention can more effectively attract users and balance the load. It can significantly improve the operating profit of charging stations in a competitive market, while taking into account the safety and stability of power grid operation and reducing wind and solar curtailment.

[0207] In summary, this invention presents a charging strategy optimization method and system considering the competitive relationship among charging stations. Based on a bounded rationality user choice model, it characterizes user risk attitudes and price sensitivity, and constructs a master-slave pricing game model to depict the decision-making behavior of electric vehicle users under uncertainty. During model solving, an adaptive ADMM is used to achieve parallel solving and rapid convergence across multiple charging stations, thereby obtaining the optimal electricity price sequence. This effectively addresses the challenges of non-convexity, high coupling, and computational complexity encountered in model solving, improving problem-solving efficiency. Using state variables such as electricity prices at different times, remaining electricity at charging stations, and the proportion of wind and solar power generation as inputs, an elastic clustering optimization model for electric vehicle charging is established within a sliding time window. This model elastically clusters and aggregates a large number of electric vehicles under different states, and introduces reinforcement learning to obtain the optimal action strategy. The charging power and charging time of electric vehicles in different clusters are dynamically allocated, thus meticulously characterizing the charging behavior of electric vehicles after elastic clustering under different states. Through the electric vehicle charging elastic clustering and charging strategy optimization module, fine-grained scheduling of charging power and time within stations is achieved under a given competitive time-of-use electricity price, thereby realizing the synergistic optimization of price and charging strategy.

[0208] There are many methods and approaches to implement the technical solution of this invention. The above description is only a preferred embodiment of this invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technology.

Claims

1.A charging strategy optimization method considering competition relationship of charging stations, characterized in that, The method comprises the following steps: A bounded rationality value model considering user risk attitude and price sensitivity is constructed to depict the selection preference of electric vehicle users among different charging stations; A master-slave game model is established for multiple charging stations with competitive relationship; An adaptive alternating direction multiplier method (ADMM) is used to iteratively solve each sub-problem in the master-slave game model to obtain the optimal time-of-use price of the charging station; Based on the optimal time-of-use price of the charging station, an electric vehicle charging elastic clustering optimization model is established in a sliding time window to aggregate and optimize a large number of vehicles in different states according to the remaining charging time, and finally the optimal charging strategy is obtained. 2.The method of claim 1, wherein, The method specifically comprises the following steps: Step 1: Obtain the total number of electric vehicles, establish an electric vehicle user charging station selection model based on bounded rationality, and determine the charging station demand function under different price strategies; Step 2: Obtain the total number of competitive charging stations, and establish a master-slave game model with the target charging station as the leader and other charging stations as the followers; Step 3: Use the sample average approximation (SAA) method to discretize the uncertainty of random arrival of electric vehicle users and distributed renewable energy output, and simulate a set of representative operating scenarios; Step 4: For the master-slave game model established in step 2, introduce the alternating direction multiplier method (ADMM) framework, and through global consensus variables and consistency constraints, decompose the original problem into multiple parallel local optimization sub-problems of charging stations; Step 5: Based on the augmented Lagrangian function, alternately update the local price strategy of each charging station, the global consensus price, and the dual variable until the algorithm converges, and obtain the optimal intraday time-of-use charging price; Step 6: Establish a charging strategy optimization model based on electric vehicle charging elastic clustering, take the intraday time-of-use charging price obtained in step 5, the remaining power of the charging station, the distributed wind and solar power generation power, and the required power of the electric vehicle as inputs, and perform elastic clustering and aggregation modeling on the electric vehicles in different states in a sliding time window to obtain the optimal charging strategy in each time period. 3.The charging strategy optimization method of considering charging station competition relationship according to claim 2, characterized in that, The electric vehicle user charging station selection model established in step S1 takes charging price, driving distance, and expected queuing time as inputs. 4.The method of claim 3, wherein, Step 1 includes: The specific method for establishing the electric vehicle user charging station selection model based on bounded rationality in step S1 includes: The total number of electric vehicles is set to ; S1-1, electric vehicle user at a charging station The overall value is expressed as: ， wherein, , and are attribute preference weights, represents a relative importance of a charging price given to each user, represents a relative importance of a travel distance given to each user, represents a relative importance of a waiting time given to each user, and ; For price deviation, , It is the user's expected price. It is the first The charging price offered by each charging station; a travel time deviation, , a travel time expected by a user, a travel time if the first electric vehicle selects the first charging station for charging; for the queuing time bias, , denotes the expected waiting time of a user, denotes the actual waiting time if the first electric vehicle chooses the first charging station for charging; represents the utility function; S1-2, using a sigmoid value function map the objective bias to subjective mental values, establishing the following equation: ， wherein, represents a property bias; is a decreasing function of the benefit sensitivity, is a loss sensitivity decreasing coefficient, , ; represents a loss avoidance coefficient, ; S1-3: The waiting time is estimated using the M / M / c queuing model: ， ， wherein, denotes the number of charging piles; denotes the average arrival rate of EV users; denotes the average service rate of a single charging pile; denotes the utilization of the charging station ; ; denotes the probability that all charging piles in the charging station are idle; denotes the number of occupied charging piles in the charging station; S1-4, the user's charging station selection behavior is simulated by a multinomial Logit The probability of the user selecting a charging station is ， wherein, denotes the set of all charging stations, denotes the user selects a charging station with utility, denotes the exponential value of Based on the probability distribution, the charging station that the user eventually selects is determined, and then the electric vehicle at the charging station The required charging energy is calculated as ， wherein, represents the SOC of the i-th electric vehicle at the current time t; represents the SOC of the i-th electric vehicle at the current time t; represents the battery capacity of the i-th electric vehicle; represents the battery capacity of the i-th electric vehicle; represents the power consumption per kilometer of the electric vehicle; represents the distance of the electric vehicle from the charging station. 5.The method of claim 4, wherein, The specific implementation method of step 2 includes: S2-1, determining a set of competitive charging stations , price upper bound , price lower bound ; setting a pricing time and a set of scenarios wherein is the total number of time periods; S2-2, Target charging station As the leader, the rest of the sites As a follower, the leader first announces the price vector. Each follower observed Then each person selects their own price. Subsequently, the user charging station selection model provides information for each station in uncertain scenarios. The following needs ,in Indicates the leader exist The price of a moment Indicates that the followers are The price of a moment; S2-3, for any follower , set up a profit maximization problem: ， ， ， wherein represents the operating cost of the competing charging station; represents the amount of electricity purchased from the grid; represents the unit cost of electricity purchased, represents the potential loss of wind energy generation, represents the potential loss of solar energy generation; and represents the distributed wind power output and the distributed photovoltaic power output; represents the electric vehicle charging power; represents the number of electric vehicles that need to be charged at the charging station at the time instant t; at the charging station. S2-4: The leader solves the profit maximization under the condition of anticipating the joint response of the followers: ， ， wherein, represents the charging demand of the electric vehicle at the time instant t. represents the charging demand of the electric vehicle at the time instant t. represents the charging demand of the electric vehicle at the time instant t. represents the charging demand of the electric vehicle at the time instant t. 6.The method of claim 5, wherein, The specific implementation method of step 3 includes: The sample average approximation SAA method is used to handle uncertainties in wind power and photovoltaic power generation. The local objective function of a charging station is expressed as: ， ， wherein, representing a scenario the lower profit of the charging station. 7.The method of claim 6, wherein, The specific implementation method of step 4 includes: S4-1, decomposing the multi-station game into a number of parallel solvable sub-problems, each sub-problem corresponding to a single charging station; S4-2, Alternating Direction Multiplier Method ADMM by introducing auxiliary variables achieve decomposition, whose core mechanism is to ensure the consistency constraints of charging stations, , denote the local decision variables of charging stations . 8.The method of claim 7, wherein, The specific implementation method of step 5 includes: S5-1, based on the objective function of equation (13) and the consistency constraint described in step 4, the charging station The augmented Lagrangian function of the charging station is: ， wherein, to constrain the associated dual variable, is a penalty parameter; S5-2, local price optimization, parallel computation update : ， Thus, the optimal price vector of each charging station for this iteration is obtained; S5-3: According to the local update results of each station, a new consensus price is calculated and the dual variable is updated: ， ， S5-4: Calculate the original residual and dual residual: ， ， S5-5, in order to accelerate the convergence speed, the adaptive penalty parameter is introduced, and the penalty parameter is dynamically adjusted according to the size of the original residual and the dual residual : ， wherein , and denote parameters, set to , ; According to formula (15), continue to update the iteration until convergence, and thus the optimal charging price of the charging station is obtained. 9.The charging strategy optimization method of considering charging station competition relationship according to claim 8, characterized in that, The specific implementation method of step 6 includes: S6-1, according to the optimal price in step 5 and the required electricity quantity for charging the electric vehicle in step 1 , according to the charging strategy at the last time, calculate the remaining charging quantity of each electric vehicle in the charging station at the current time , and set the sliding time ; S6-2, at the current time, according to the charging station state, the time required for charging and the remaining time of all electric vehicles in the charging station are compared, and the electric vehicles at the current time are aggregated and classified: ， ， wherein, denotes the rated charging power of the charging station, denotes the aggregated classification, denotes the required remaining charging time of the electric vehicle, denotes the ceiling function; S6-3, for each electric vehicle in the cluster, the following equation is established to determine the surplus time of the electric vehicle in the cluster, and the electric vehicle with less surplus time has higher charging priority: ， wherein, represents the surplus time of each electric vehicle in the th aggregation. S6-4, for the aggregated classification, set the action sequence , the action sequence represents the charging probability of the whole electric vehicle in each aggregation, wherein , ; According to the current time state , respectively, the occupation proportion of charging piles in the charging station and the occupation proportion of wind and light power generation, the wind and light proportion and the profit maximization problem are established, that is, the objective function, , the established objective function is as follows: ， ， wherein, represents a weight coefficient, and is a number between 0 and 1; represents an action policy; represents the number of electric vehicles in the kth aggregation at the current time. S6-5, using Q-learning in The following Q value iteration formula is established in the above equation, and the behavior with the maximum Q value is obtained as the optimal charging behavior through iterative learning according to the current state. ， wherein, is a learning rate, is a discount factor, denotes the total number of time periods; S6-6, according to formula (23), the electric vehicle priority is determined, the number of electric vehicles selected for charging in each cluster is determined according to the optimal charging behavior and priority of the cluster, and then the optimal charging strategy is obtained by multiplying the charging quantity by the charging power. 10.A charging strategy optimization system considering competition relationship of charging stations, characterized in that, The system comprises: A bounded rationality value model construction module is configured to construct a bounded rationality value model considering user risk attitude and price sensitivity, and to depict the selection preference of electric vehicle users among different charging stations; A game model construction module is configured to establish a master-slave game pricing model for a plurality of charging stations in competitive relationship; An optimal time-of-use electricity price acquisition module is configured to solve each sub-problem in the master-slave game model by using an adaptive alternating direction multiplier method (ADMM) iteration to obtain an optimal time-of-use electricity price of the charging station; An optimal charging strategy acquisition module is configured to obtain the optimal time-of-use electricity price of the charging station based on the optimal time-of-use electricity price obtained by the optimal time-of-use electricity price acquisition module, to aggregate and optimize a large number of vehicles in different states according to the remaining charging time by establishing an electric vehicle charging elastic clustering optimization model in a sliding time window, and to finally obtain an optimal charging strategy. 11.The charging strategy optimization system considering charging station competition relationship according to claim 10, wherein, The system further comprises an initialization module configured to input competitive pricing parameters. 12.The charging strategy optimization system considering charging station competition relationship according to claim 11, wherein, The system further comprises: A constraint condition construction module is configured to construct a price upper limit, a price lower limit, a queuing stability constraint, and an energy balance constraint according to charging pile quantity, charging power upper limit, wind and solar output, power grid purchasing capacity, and queue stability requirements, so as to limit the feasible solution space of the charging station pricing game model; An optimization objective function construction module is configured to set an objective function of charging station pricing optimization based on the constraint conditions, and specifically to maximize charging station operating profit as the charging station time-of-use pricing optimization objective by comprehensively considering purchasing cost, renewable energy generation loss, and charging service revenue; A scenario simulation module is configured to use a sample average approximation method to perform scenario division and simulation on random arrival behavior of electric vehicle users and uncertainty of distributed wind power and photovoltaic power generation, and to generate a set of representative operation scenarios to provide data support for subsequent competitive game strategy evaluation; A competitive game strategy screening module is configured to compare and screen different competitive pricing strategies based on demand response of each candidate time-of-use electricity price strategy obtained by scenario simulation under the master-slave game framework, and to output a charging station competitive pricing strategy that meets the constraint conditions and optimizes the objective function.

Images