A method and system for guiding and scheduling charging load of shared electric vehicles

Abstract

The application discloses a shared electric vehicle charging load guidance scheduling method and system, establishes a charging load guidance scheduling game model of an upper power traffic coupling system operator and a lower shared electric vehicle fleet based on Stackelberg-Nash game, determines a target function and constraint conditions of the charging load guidance scheduling game model, the target function includes a first target function of minimum cost of the shared electric vehicle fleet considering charging and discharging incentives and rebalancing incentives, and a second target function of minimum cost of the power traffic coupling system operator considering incentive response, and the charging load guidance scheduling scheme is obtained by using a super gradient descent algorithm to distributively solve the charging load guidance scheduling game model. In this way, the optimal traffic and power grid service incentive strategy of the upper system operator is guided by the scheduling model, the lower fleet is guided and the charging load is scheduled, and the collaborative optimization of energy supply and demand and traffic flow distribution is achieved.

Description

Technical Field

[0001] This invention relates to the technical field of power system dispatching, and in particular to a method and system for guiding and dispatching charging loads for shared electric vehicles. Background Technology

[0002] With the widespread application of shared electric vehicles (SEVs) in intelligent transportation systems and smart grids, the coordinated optimization of vehicle-to-grid (V2G) interaction and rebalancing scheduling has become an important means to improve system efficiency. SEVs can not only feed back electrical energy to the grid through V2G, providing flexible power support, but also optimize vehicle spatial distribution through reasonable rebalancing scheduling, thereby alleviating urban traffic congestion.

[0003] However, in actual operation, when SEV fleets make charging and discharging decisions and rebalancing choices, they are not only affected by electricity prices, but also need to take into account factors such as travel costs, route preferences, charging facility capacity limitations, and rebalancing incentives.

[0004] Therefore, how to design a reasonable incentive mechanism to guide the SEV fleet to optimize its own behavior, while realizing the collaborative interaction between the electric-transportation coupling system operator (CSO) and the SEV fleet, is an urgent problem to be solved. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a charging load guidance and scheduling method and system for shared electric vehicles, which fully integrates the two dimensions of charging and discharging incentives and rebalancing incentives. It aims to guide the fleet to make reasonable route selection and rebalancing decisions on the basis of ensuring the stable operation of the power system, thereby achieving coordinated optimization of energy supply and demand and traffic flow distribution.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for guiding and scheduling charging loads for shared electric vehicles, comprising the following steps: Acquire electric transportation coupling system operators and shared electric vehicle fleets; Establish a charging load guidance and scheduling game model based on Stackelberg-Nash game theory between upper-level electric transportation coupling system operators and lower-level shared electric vehicle fleets; The objective function and its constraints of the charging load guidance and scheduling game model are determined. The objective function includes a first objective function that considers the minimum cost of the shared electric vehicle fleet considering charging and discharging incentives and rebalancing incentives, and a second objective function that considers the minimum cost of the electric transportation coupling system operator considering incentive response. The constraints include a first constraint corresponding to the first objective function and a second constraint corresponding to the second objective function. The charging load guidance and scheduling game model is solved in a distributed manner using the supergradient descent algorithm to obtain the charging load guidance and scheduling scheme.

[0007] To solve the above-mentioned technical problems, another technical solution adopted by the present invention is as follows: A charging load guidance and scheduling system for shared electric vehicles includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the various steps of the aforementioned charging load guidance and scheduling method for shared electric vehicles.

[0008] The beneficial effects of this invention are as follows: It establishes a Stackelberg-Nash game-based charging load guidance and scheduling game model for upper-level electric transportation coupling system operators and lower-level shared electric vehicle fleets; it determines the objective function and constraints of the charging load guidance and scheduling game model, including a first objective function considering the minimum cost of the shared electric vehicle fleet based on charging / discharging incentives and rebalancing incentives, and a second objective function considering the minimum cost of the electric transportation coupling system operator based on incentive responses; it uses a distributed solution of the charging load guidance and scheduling game model using a super-gradient descent algorithm to obtain a charging load guidance and scheduling scheme. In this way, the scheduling model guides upper-level system operators to formulate optimal traffic and power grid service incentive strategies, achieving guidance for the lower-level fleet and scheduling of charging loads, thus achieving coordinated optimization of energy supply and demand and traffic flow distribution. Attached Figure Description

[0009] Figure 1 This is a flowchart of a charging load guidance and scheduling method for shared electric vehicles according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a charging load guidance and scheduling system for shared electric vehicles according to an embodiment of the present invention; Figure 3 This is a diagram illustrating the collaborative scheduling framework for shared electric vehicles under the coupling of electric and transportation systems, as described in an embodiment of the present invention. Figure 4 This is a diagram illustrating the information interaction framework of the Stackelberg-Nash game according to an embodiment of the present invention. Figure 5 This is a system diagram of a computational example according to an embodiment of the present invention; Figure 6This is a revenue iteration trend diagram for the operator of the electric transportation coupling system according to an embodiment of the present invention; Figure 7 This is a traffic congestion level diagram that only considers vehicle-to-network interaction incentives in an embodiment of the present invention; Figure 8 A traffic congestion map that simultaneously considers vehicle-to-network interaction incentives and rebalancing incentives in an embodiment of the present invention; Figure 9 This is a comparison chart of the revenue of a shared electric vehicle fleet under different budgets according to an embodiment of the present invention; Figure 10 This is a comparison chart of the revenue of shared electric vehicle fleets at different scales according to an embodiment of the present invention; Figure 11 This is a convergence trend diagram of the excitation of vehicle convoy No. 1 in an embodiment of the present invention; Figure 12 This is a convergence result diagram of the sensitivity of vehicle convoy No. 1 in an embodiment of the present invention; Figure 13 This is a convergence trend diagram of the decision variables for vehicle 1 in an embodiment of the present invention.

[0010] Label Explanation: 1. A charging load guidance and scheduling system for shared electric vehicles; 2. Memory; 3. Processor. Detailed Implementation

[0011] To explain in detail the technical content, objectives, and effects of the present invention, the following description is provided in conjunction with the embodiments and accompanying drawings.

[0012] In existing technologies, there are problems such as the incentive mechanism design in electric vehicle scheduling not fully integrating the dual incentives of transportation and power grid, and the lack of personalized incentive design.

[0013] To address at least the aforementioned technical problems, this invention provides a method for guiding and scheduling charging loads for shared electric vehicles, comprising the following steps: Acquire electric transportation coupling system operators and shared electric vehicle fleets; Establish a charging load guidance and scheduling game model based on Stackelberg-Nash game theory between upper-level electric transportation coupling system operators and lower-level shared electric vehicle fleets; The objective function and its constraints of the charging load guidance and scheduling game model are determined. The objective function includes a first objective function that considers the minimum cost of the shared electric vehicle fleet considering charging and discharging incentives and rebalancing incentives, and a second objective function that considers the minimum cost of the electric transportation coupling system operator considering incentive response. The constraints include a first constraint corresponding to the first objective function and a second constraint corresponding to the second objective function. The charging load guidance and scheduling game model is solved in a distributed manner using the supergradient descent algorithm to obtain the charging load guidance and scheduling scheme.

[0014] In this way, embodiments of the present invention can guide upper-level system operators to formulate optimal traffic and power grid service incentive strategies through a scheduling model, thereby guiding the lower-level vehicle fleet and scheduling the charging load, achieving coordinated optimization of energy supply and demand and traffic flow distribution.

[0015] In the following, the technical solutions according to this disclosure will be described with reference to specific embodiments and in conjunction with the accompanying drawings.

[0016] Please refer to Figure 1 and Figure 3 Embodiment 1 of the present invention is as follows: Figure 1 This is a flowchart illustrating a charging load guidance and scheduling method for shared electric vehicles according to an embodiment of the present disclosure. (Refer to...) Figure 1 Method 1 includes steps 101 to 104.

[0017] In step 101, the electric transportation coupling system operator and the shared electric vehicle fleet are obtained.

[0018] In step 102, a charging load guidance and scheduling game model based on Stackelberg-Nash game is established between the upper-level electric transportation coupling system operator and the lower-level shared electric vehicle fleet.

[0019] In some embodiments, the charging load guidance and scheduling game model is as follows:

[0020] In the formula, f i Let Ω represent the first objective function. i This indicates the first constraint condition. Let Ω represent the second objective function. CSO This represents the second constraint, N represents the shared electric vehicle fleet, and CSO represents the electric transportation coupling system operator.

[0021] In this way, the equilibrium solution of the charging load-guided scheduling game model can balance the behavior of electric transportation coupling system operators and shared electric vehicle fleets, aiming to achieve a win-win situation rather than just improving individual interests.

[0022] In some embodiments, the game theory model for charging load guidance scheduling includes the following steps: P1, the electric transportation coupling system operator broadcasts an incentive to all shared electric vehicle fleets; P2, the shared electric vehicle fleets determine their optimal response strategy based on the incentive broadcast by the electric transportation coupling system operator, combined with a first objective function and a first constraint; P3, the electric transportation coupling system operator determines the optimal incentive based on the optimal response strategies of all shared electric vehicle fleets, combined with a second objective function and a second constraint; P1 to P3 are repeated until the determined optimal incentive remains unchanged. In this way, the interaction between the electric transportation coupling system operator and multiple shared electric vehicle fleets during the scheduling process is captured.

[0023] In step 103, the objective function and its constraints of the charging load guidance and scheduling game model are determined. The objective function includes a first objective function that considers the minimum cost of the shared electric vehicle fleet considering charging and discharging incentives and rebalancing incentives, and a second objective function that considers the minimum cost of the electric transportation coupling system operator considering incentive response. The constraints include a first constraint corresponding to the first objective function and a second constraint corresponding to the second objective function.

[0024] In some embodiments, a first revenue value for the shared electric vehicle fleet to provide electricity is calculated considering charging and discharging incentives; a second revenue value for the shared electric vehicle fleet to provide rebalancing is calculated considering rebalancing incentives; the deviation cost of the shared electric vehicle fleet changing routes or destinations is calculated; the travel time cost of the shared electric vehicle fleet is calculated; the sum of the deviation cost and the travel time cost is subtracted from the sum of the first and second revenue values ​​to obtain the aggregate cost of the shared electric vehicle fleet; and the first objective function is obtained by minimizing the aggregate cost of the shared electric vehicle fleet. The first constraints include traffic flow conservation constraints, upper and lower bound constraints on the decision vector of the first objective function, and coupling resource constraints. In this way, the cost function of each fleet considers four key factors: the revenue from providing energy or rebalancing services, travel time, and travel preferences. Each fleet aims to minimize its individual cost by selecting the probabilities of route selection and destination selection.

[0025] In some embodiments, the formula for calculating the first revenue value is as follows:

[0026] In the formula, This represents the first profit value. Indicates the location of the charging station. Indicates the electric transportation coupling system operator at the destination d For the team i Excitation of discharge, This indicates the degradation loss of the battery during the discharge process. Indicates the firsti The team selected the first d The probability of a destination. m i Indicates the team i The number of shared electric vehicles owned; in this way, incentives can be designed to guide SEVs to perform energy services or redistribute them to specific locations, such as specific charging stations or rebalancing areas.

[0027] The formula for calculating the second revenue value is:

[0028] In the formula, This represents the second revenue value. Indicates the rebalancing site. Indicates the electric transportation coupling system operator at the destination d For the team i The provided rebalancing incentives This indicates the degradation loss of the battery during the rebalancing process.

[0029] In some embodiments, the total travel time cost of all shared electric vehicle fleets is calculated; the discharge revenue of the electric transportation coupling system operator is calculated; the rebalancing revenue of the electric transportation coupling system operator is calculated; the total travel time cost is subtracted from the sum of the discharge revenue and the rebalancing revenue to obtain the cost of the electric transportation coupling system operator; and the second objective function is obtained by minimizing the cost of the electric transportation coupling system operator. The second constraints include upper and lower bound constraints on the decision vector of the second objective function and constraints on the total amount of incentive resources. In this way, the cost function of the CSO includes total travel time, total discharge revenue, and total rebalancing revenue, minimizing energy supply costs to avoid exacerbating traffic congestion.

[0030] In some embodiments, the total incentive resource constraint includes:

[0031] In the formula, C b Indicates the total amount of incentives. C Indicates the upper limit of the total incentive amount. p It indicates motivation. A vector representing the traffic flow to the destination; Indicates the location of the charging station. This represents the excitation from the electric transportation coupling system operator to the vehicle fleet i at destination d. This indicates the degradation loss of the battery during the discharge process; Indicates the rebalancing site. This represents the rebalancing incentive provided by the electric transportation coupling system operator to fleet i at destination d. This indicates the degradation loss of the battery during the rebalancing process; This indicates the assembly of the convoy. m i Indicates the team i The number of shared electric vehicles owned. This approach effectively considers the limited nature of incentive resources. Indicates the team i The total flow rate reaching each charging station.

[0032] In step 104, the charging load guidance and scheduling game model is solved in a distributed manner using the supergradient descent algorithm to obtain the charging load guidance and scheduling scheme. In this way, the distributed solution method based on supergradient descent can protect fleet privacy and improve computational efficiency.

[0033] As can be seen, this embodiment proposes a shared electric vehicle scheduling method that considers personalized incentive design. This method is based on the Stackelberg-Nash game model to characterize the hierarchical interaction between the electric transportation coupling system operator and the SEV fleet: at the upper level, the system operator formulates traffic and grid service incentives under the constraint of limited incentive budget, guiding the rebalancing and V2G decision-making behavior of the lower-level SEV fleet; at the lower level, Nash game is used to describe the non-cooperative game relationship between multiple SEV fleets.

[0034] This model primarily comprises two types of game processes: Stackelberg game and Nash game. Stackelberg SO, acting as the leader, guides the charging and discharging behavior and travel decisions of the SEV fleet by jointly setting V2G and rebalancing incentives, optimizing the allocation of power and transportation resources from a system-wide perspective. Nash game, where SEV fleet managers, acting as followers, make route selection and charging / discharging decisions based on their own profit maximization under the V2G and rebalancing incentives set by the CSO, competing for limited V2G equipment and road resources within the system. Within this game framework, the CSO optimizes system operating costs through a dual incentive mechanism (V2G incentives and rebalancing incentives), improving scheduling efficiency and resource utilization while ensuring service quality and user experience.

[0035] System framework such as Figure 3 As shown, the electric-transit coupling system operator sends incentive information to the shared electric vehicle fleet, which in turn returns trip selection information and incentive response sensitivity to the electric-transit coupling system operator. The shared electric vehicle fleet can also send traffic flow and energy flow to coupling resources (including charging station resources and road resources), and obtain congestion information from the coupling resources.

[0036] Due to the large number of SEV fleets and the highly decentralized individual decision-making, traditional centralized optimization methods face high computational costs and privacy risks in this context. To address this challenge, this embodiment employs a distributed solution method based on the hypergradient descent algorithm, enabling the CSO and SEV fleets to perform interactive optimization in a distributed environment, ultimately converging to the SNG equilibrium solution. Compared to centralized solution methods, this approach not only significantly reduces computational complexity and protects fleet privacy but also possesses good scalability, making it suitable for large-scale SEV scheduling scenarios and improving the system's feasibility and practical deployment value.

[0037] Please refer to Figure 4 Embodiment two of the present invention is as follows: Based on Embodiment 1, this embodiment provides a specific implementation step for a charging load guidance and scheduling method for shared electric vehicles, including steps 201 to 204: Step 201: Consider a shared electric vehicle scheduling model with dual excitation of charging / discharging and rebalancing.

[0038] Specifically, consider a CSO that manages multiple SEV fleets through a well-designed incentive mechanism. Each fleet manager is responsible for dispatching a group of SEVs, which can choose to provide travel services based on actual needs, or provide V2G power support to the grid in the event of a power shortage.

[0039] In this embodiment, the traffic network is modeled as a connected graph G. Where G represents the traffic network topology. Represents the set of traffic nodes. This represents the set of road segments connecting these nodes. Charging stations and rebalancing locations correspond to subsets of the node set, denoted as... and It should be noted that, to avoid functional overlap, the following is set... This means that each node can only serve as either a charging station or a rebalancing location. The destination set is defined as follows: It contains Each element.

[0040] Step 2011: Establish the objective function for the travel cost of the shared electric vehicle fleet.

[0041] Specifically, each team i The aim is to select probabilities by choosing road segments. v i and the probability of destination selection w i To minimize individual costs and maximize individual benefits. Each team i The cost function considers four key factors: the revenue from providing energy or rebalancing services, travel time, and travel preferences. First, to incentivize SEVs to provide energy services or redistribute energy to specific locations, incentives should be designed for specific charging stations or rebalancing areas. Revenue generated from providing electricity at target nodes should be considered. This can be described using the following function:

[0042] In the formula, Indicates the location of the charging station. Indicates the first i The team selected the first d The probability of a destination. Indicates the electric transportation coupling system operator at the destination d For the team i Excitation of discharge, This indicates the degradation loss of the battery during the discharge process. m i Indicates the team i The number of shared electric vehicles owned.

[0043] Similarly, the team i The return on reaching a certain rebalancing region It is expressed as follows:

[0044] In the formula, Indicates the rebalancing site. Indicates the electric transportation coupling system operator at the destination d For the team i The provided rebalancing incentives This indicates the degradation loss of the battery during the rebalancing process.

[0045] Then, the SEV fleet typically has pre-planned routes and destinations during daily operations. When the CSO issues incentives, the SEV managers are attracted and may change the operating routes or destinations, thus creating a deviation cost. ,as follows:

[0046] In the formula, This indicates a team-specific preference parameter. A larger size means the convoy will place greater emphasis on the destination. A larger size means that the team will pay more attention to the route; The routes and destinations preferred by the fleet manager are defined as follows:

[0047] In the formula, Indicates the first iThe team was assigned to the first e The basic vehicle ratio for each road segment, where E represents the number of road segments; Indicates the first i The team selected the first d The base probability of a destination D Indicates the number of destinations.

[0048] Finally, due to road capacity limitations, too many vehicles choosing the same route will cause congestion, resulting in negative congestion time costs. i Expected travel time cost The definition is as follows:

[0049] In the formula, Indicates the team i unit time cost t e Indicates the calculated road segment e Travel time, This indicates the calculation of total traffic flow. v e This indicates that each team selects the first... e A vector composed of the proportion of vehicles on each road segment Indicates the first i The team was assigned to the first e The proportion of vehicles on each road segment.

[0050]

[0051]

[0052] In the formula, Indicates road segment e A positive coefficient for the degree of congestion. Indicates road segment e The total number of other vehicles on the road.

[0053] Therefore, the SEV team i The payoff function consists of four parts: discharge payoff Rebalancing income Costs associated with deviations from the desired route and destination and travel time revenue Among them, travel time revenue It is influenced not only by the choices of individual teams, but also by the decisions of other teams. The remaining three components are influenced by their own decisions and the incentives provided by the CSO.

[0054] SEV Team i The aggregation cost function is as follows:

[0055] In the formula, v The matrix representing the route selections of each vehicle team. w The matrix representing the destination choices of each convoy. σ ( v ) represents a vector consisting of traffic flow on each road.

[0056] Step 2012: Establish constraints, including flow conservation constraints, upper and lower bound constraints for decision variables, and coupled resource constraints.

[0057] Traffic flow conservation constraint: To ensure the conservation of traffic flow in the transportation network, that is, for each road segment... e In this regard, the inflow of traffic must equal the outflow, thus defining the following constraint (1.1):

[0058] In the formula, Indicates from j Node to k The road segment at the node, Indicates from k Node to m The road segment at the node, d Indicates the destination number. Indicates the first i The team selected the first d The probability of a destination. o i This represents the starting point of the i-th convoy.

[0059] Upper and lower bound constraints on decision variables: fleet i The selection of routes and destinations is modeled using probability distributions, therefore the following constraints (1.2) must be satisfied:

[0060]

[0061] Coupled resource constraints: To ensure that the traffic flow to each destination does not exceed the upper limit, the following constraint (1.3) is introduced:

[0062] In the formula, Indicate destination d The maximum number of vehicles that can be accommodated. Indicates road segment e The maximum number of vehicles that can be accommodated.

[0063] Define Ω iIndicates the team i Decision variables Feasible areas:

[0064] Step 202, consider the operator cost model of the electric-transportation coupled system with incentive response.

[0065] Specifically, the CSO is responsible for designing personalized incentives to encourage SEV fleets to perform V2G feeding or rebalancing at charging station nodes and designated rebalancing areas within the transportation network. In particular, the CSO assigns specific incentives to each fleet. i At each destination d Different incentive strategies are formulated. The goal of this strategy is to maximize energy supply benefits while avoiding exacerbating traffic congestion. Therefore, the revenue function of a CSO includes total travel time, total discharge revenue, and total rebalancing revenue.

[0066] Step 2021: Consider the multi-objective cost function with limited incentive resources.

[0067] The cost function of a CSO is as follows:

[0068] In the formula, This represents the total travel time cost. Indicates discharge benefit, This indicates the rebalancing return.

[0069] The total travel time cost of the convoy across all road segments is calculated by adding up the travel times of each convoy:

[0070] In the formula, , μ This represents the revenue per unit time for the CSO (Congestion and Speed ​​Occupancy). Total travel time is a key indicator for assessing traffic congestion and the overall efficiency of the transportation network. A negative sign indicates that the longer the total travel time, the greater the negative impact.

[0071] CSOs can benefit from the electrical energy provided by SEVs, relying on a fleet of SEVs to supply power to critical load nodes. The discharge benefits take the following forms:

[0072] In the formula, Indicate destination d Traffic flow vector;

[0073] This indicates that each convoy has arrived at the charging station. dThe total traffic flow β 1. β 2 represents the V2G revenue parameter of CSO. This quadratic form indicates that the marginal revenue from discharge should not increase with the increase in the number of vehicles.

[0074] The rebalancing process of SEVs reduces traffic congestion and helps stabilize the efficiency of the transportation system. From a system perspective, the rebalancing benefits can be expressed as a linear function:

[0075] In the formula, β 3 represents the rebalancing revenue parameter of CSO.

[0076] Step 2022: Establish constraints, including boundary constraints on decision variables and total incentive resource constraints.

[0077] Specifically, the CSO at the destination d For SEV team i The set incentive value, intended to guide fleet behavior, must be positive, but cannot exceed the maximum incentive value that the CSO can provide. Therefore, the following constraint (2.1) is introduced:

[0078] In the formula, This indicates that the CSO is at the destination. d Provides the maximum value of the discharge excitation. This indicates that the CSO is at the destination. d Provides the maximum value of the rebalancing stimulus.

[0079] The CSO distributes incentives to the SEV fleet based on a pre-set budget. Therefore, the following incentive cap constraint must be met:

[0080] In the formula, C b Indicates the total amount of incentives. C Indicates the upper limit of the total incentive amount. p Indicates motivation;

[0081] Indicates the team i The total flow rate reaching each charging station.

[0082] The incentive cap constraint is reformulated as a soft constraint and incorporated into the cost function of the CSO for easier calculation, resulting in:

[0083] Accordingly, the revenue function of the merged CSO takes the following form:

[0084] In the formula, This represents the penalty coefficient used to ensure compliance with the incentive cap constraint.

[0085]

[0086] In the formula: Represents a personalized incentive vector. Let represent the set of D×N dimensional real matrices.

[0087] Step 203: Analysis and solution of the scheduling model based on Stackelberg-Nash game.

[0088] Specifically, a Stackelberg-Nash game model is constructed to capture the interactions among the participants in the scheduling process. At the upper level, the CSO, as the leader, is responsible for making incentive decisions; at the lower level... N Each SEV fleet is considered a follower and responds to incentives from the CSO. The CSO broadcasts personalized incentive information to each fleet, which the fleet responds by choosing to be reassigned to a rebalancing zone or to perform V2G services. This response process can be determined by solving a set of interdependent optimization problems that constitute the underlying Nash game.

[0089] Therefore, the SNG model with a leader and N followers is expressed as follows:

[0090] The model consists of the following parts: Participants: CSO and Collection SEV fleet; Strategy set: Incentives set by CSO p Travel options with all fleet agents Cost function: for SEV fleet The travel cost function is defined in step 2011, and the total system cost function of the CSO is defined in step 2021.

[0091] The game theory model in this embodiment assumes that all participants are rational and self-interested, meaning they understand the current situation, identify all possible scenarios, and maximize their expected gains. By gradually eliminating strictly dominant strategies, these rational participants will eventually reach a Nash equilibrium. Therefore, the equilibrium solution of the proposed SNG model can balance the behavior of all participants, aiming to achieve a win-win situation, rather than simply increasing individual interests.

[0092] Step 2031, Stackelberg-Nash game process.

[0093] Please refer to Figure 4 In a standardized Stackelberg-Nash game with a single leader and N followers, the process is as follows: P1. The leader selects a strategy from their strategy set and broadcasts it to all followers. Given the leader's decision priorities, when the leader broadcasts their strategy to all followers, it is considered an incentive. p At that moment, the game began.

[0094] P2. Followers determine their optimal response strategy based on the leader's strategy. Specifically, after receiving the leader's strategy, the followers engage in a non-cooperative Nash game to reach a Nash equilibrium (NE), which is obtained as follows:

[0095] In the formula, This represents the decisions of all other teams in the lower level except for team i.

[0096] P3. The leader determines its optimal strategy based on the best response strategies of all followers. The optimal response strategy calculated in step two is used. Leaders then select the best strategy from their set of feasible strategies. The calculation is as follows:

[0097] P4. Repeat steps P1 to P3 to find the best strategy for the leader and followers until all participants no longer change their strategies, indicating that the optimal strategy has been found.

[0098] Step 2032, Stackelberg-Nash game equilibrium analysis.

[0099] In game theory, the solution to a Nash game is typically defined as a Nash equilibrium (NE), where no player can unilaterally deviate from the NE to increase their payoff. In the model of this example, the NE describes the equilibrium at the lower level. Similarly, the solution to a Stackelberg game is typically defined as a Stackelberg equilibrium (SE), where no player can unilaterally change their strategy; the SE describes the balance between the upper and lower levels. Based on these definitions of equilibrium, the concept of a solution to a Stackelberg-Nash game is represented using a Stackelberg-Nash equilibrium (SNE), such as... Figure 4 As shown.

[0100] According to the SNG model defined above, if and only if At that time, a set of strategies To constitute an SNE, the following conditions must be met:

[0101] In the formula, This indicates the optimal response of the lower-level vehicle i. This represents the optimal response of all other followers to the leader's strategy; the first inequality in the equation indicates that the leader cannot achieve this through choice. Other strategies to further increase their income; the second inequality indicates that the leader chooses the optimal strategy After that, no follower can change their strategy unilaterally. This increases their revenue. In SNE, both CSO and SEV teams maximize their profits.

[0102] To address this equilibrium, existing literature typically reformulates the entire problem as a mixed-integer programming problem and solves it using existing software to find the global optimum. However, as the problem size increases, the computational complexity rises sharply, making it unsuitable for large-scale problems. Therefore, step 2032 will employ a hypergradient algorithm, focusing on efficiently finding local optima.

[0103] Step 2032, Distributed solution method based on supergradient descent algorithm.

[0104] This embodiment employs a solution method based on the supergradient descent algorithm, which is designed to be extended to larger-scale two-level game problems. The algorithm is based on weak assumptions about the upper-level objective function, adopts a general form, and is optimized using a simple and easily implemented iterative update rule. Furthermore, this method preserves the distributed structure of the game problem, allowing the lower level to accommodate a large number of participants, thus exhibiting good scalability.

[0105] Specifically, define a from Single-valued mapping to NE:

[0106] And this mapping is used to construct the optimization problem of CSO, as shown below:

[0107] CSO objective function Depends on This indicates that during the optimization process, the CSO needs to predict the SEV fleet's performance. A rational response. Regarding The derivative form is expressed as:

[0108] The core objective of this algorithm is to find local optima in optimization problems using the projective gradient descent method. This process can be characterized by the following chain rule:

[0109] The outer loop SNE solution algorithm of this algorithm is shown in the following pseudocode: Parameter: Step size threshold ; initialization: , , , ; Iterate until convergence: For CSO: Calculate the pseudo gradient:

[0110] Update incentives:

[0111] Will The announcement was broadcast to the SEV team manager. For CSO + SEV fleet managers: Estimating route selection and sensitivity:

[0112] Each iteration .

[0113] Regarding the pseudocode above, in order to calculate the upper-level gradient, CSO needs to obtain... and its Jacobian matrix, denoted as This Jacobian matrix is ​​used to describe the sensitivity of the lower-level SEV team, and can intuitively reflect the lower-level SEV team's response to incentives. The response. From a mathematical perspective, the gradient descent algorithm calculates a quantity called the "hypergradation," denoted as... It is a generalization of non-smooth and non-convex functions.

[0114] This two-level game model employs a double-loop iterative approach to achieve hierarchical decision-making solutions. In each outer loop iteration... In the middle, CSO presented to the SEV team Broadcast the current personalized incentives. Subsequently, in each inner loop iteration, the SEV fleet manager aggregates the variables based on the CSO broadcast. and The selection of its driving route and destination Estimate and update their responses to stimuli, i.e., their sensitivity. .

[0115] The inner loop NE solution algorithm is shown in the following pseudocode: Parameter: Step size γ ; enter: p , x , s , σ ; definition:

[0116] initialization: , , ,

[0117] , ,

[0118] Iterate until convergence: For SEV team managers ( *(Parallel): Update path selection:

[0119] when When =1, update , and :

[0120]

[0121]

[0122] when If the value is ≠ 1, do not update. , and .

[0123] Update sensitivity:

[0124] For CSO: polymerization:

[0125] like ,but =1, otherwise, =0; broadcast ; Each iteration until ; Final output .

[0126] Regarding the pseudocode above, the inner loop terminates once the estimation result reaches sufficient accuracy. Although the outer loop does not need to... and An exact solution is obtained, but the CSO still collects approximate NEs and their sensitivities, i.e. and And use this information to update the activation through projected gradient descent. .in, This represents the inaccurate gradient obtained from the calculation, i.e., the hypergradient.

[0127] In the inner-loop NE solution algorithm, in order to update traffic flow, the SEV fleet manager's decision needs to be projected onto a set of polyhedra. This is equivalent to repeatedly solving a parameterized quadratic programming problem, which can be solved efficiently with a suitable solver. On the other hand, the feasible solution set of CSO is a closed set, so the process of projecting onto this set can be computed analytically. The inner loop NE solution algorithm updates the sensitivity by requiring each fleet manager i to compute its corresponding auxiliary matrix. , and These matrices record the mapping about , , The Jacobian matrix.

[0128] Please refer to Figures 5 to 13 Embodiment 3 of the present invention is as follows: This embodiment is based on Embodiment 1 or Embodiment 2, and provides a computational example analysis: This embodiment was analyzed using the IEEE 33-node system and the improved Nguyen Dupuis traffic system (31 road segments and 22 nodes), such as Figure 5 As shown in the diagram, this electric-transport coupled system includes two charging stations and two rebalancing locations. The charging stations are located at the nodes where the electric and transportation networks are coupled. The SEV fleet departs from node 1 and travels to its destinations at nodes 13, 20, 22, and 24.

[0129] In the lowest-cost competitive model, the CSO broadcasts the electricity purchase price and rebalancing incentives to the SEV fleet managers. p The SEV team then selects its route. x and response to price and incentives sProvide feedback to the CSO. Each SEV fleet has destination and route preferences; some fleets prioritize destinations, while others focus on route selection. Therefore, this embodiment considers scheduling 10 SEV fleets, each with different preference parameters. α 1 and α 2. The available electrical energy for each SEV fleet follows a uniform distribution: P ~ N (25,2), the number of dispatchable vehicles varies from 10 to 15. In addition, the unit time revenue constants for CSO and SEV fleets are set to 0.01 and 1, respectively.

[0130] To further verify the effectiveness and feasibility of the proposed two-layer game model and solution algorithm in practical operation, this embodiment conducts numerical verification from multiple perspectives. First, by analyzing the convergence of the total payoff of the CSO, its ability to optimize the overall system payoff during the game scheduling process is evaluated.

[0131] (1) Verification of the iterative trend of total revenue of CSO In the two-level game structure, the CSO, as the upper-level leader, aims to improve the overall system revenue by designing personalized electricity price incentives to guide the SEV fleet to participate rationally in V2G discharge and rebalancing. The achievement of this objective is verified by tracking and analyzing the changing trend of the CSO's total revenue during algorithm iterations. Figure 6 As shown, the components of the CSO revenue function converged essentially by the 21st iteration and tended to stabilize in subsequent iterations. The total revenue fluctuated throughout the iteration process, eventually stabilizing at a positive level, indicating that the revenues of the CSO and SEV fleets gradually reached an equilibrium during the optimization process.

[0132] This positive return is primarily due to the guiding role of the incentive mechanism. The CSO effectively mobilizes the SEV fleet to participate in V2G discharge and rebalancing through incentives; these two components constitute the core source of the total revenue. In the 8th and 20th iterations, the penalty cost term showed non-zero values, indicating that the incentive expenditure at that stage exceeded the set budget limit. At this point, the algorithm automatically adjusts its strategy and continues to iterate to ensure that the incentive limit constraint is met.

[0133] Ultimately, the algorithm converges to a set of optimal incentive schemes that satisfy the budget constraints, ensuring sufficient V2G energy revenue in the system while effectively rebalancing traffic flow through incentive guidance. This achieves the comprehensive goals of mitigating power outage losses, improving energy efficiency, and maintaining the stable operation of the transportation system.

[0134] (2) Verification of the effect of rebalancing incentives on alleviating traffic congestion To further verify the role of rebalancing incentives in traffic guidance, this embodiment will compare and analyze the differences in system performance with and without the introduction of rebalancing incentives, focusing on the congestion level of the traffic network. Figure 7 and Figure 8 As shown in the diagram, in the scheme without rebalancing incentives, the CSO only considers guiding the SEV fleet to participate in V2G discharge through electricity price incentives, without regulating the vehicle path distribution and spatial rebalancing behavior. Simulation results show that in this case, significant traffic aggregation occurs near some nodes, and the load on some paths approaches the capacity limit, resulting in a significant risk of congestion. Due to the lack of balanced guidance, the SEV fleet tends to concentrate on the same high-efficiency destination, leading to an imbalance in traffic load distribution and thus affecting the overall operating efficiency of the system. In contrast, after introducing rebalancing incentives, the CSO considers both discharge benefits and traffic distribution effects when formulating incentive strategies, guiding some SEV fleets to the rebalancing area and achieving orderly adjustment of vehicle spatial distribution. From the heat map and path load statistics, it can be seen that the path flow in the traffic network is more balanced, the number of high-load paths is reduced, and the overall congestion is alleviated.

[0135] This comparative verification demonstrates that rebalancing incentives, as a key control mechanism in the game theory model, can effectively guide convoys to avoid highly congested areas, thereby enhancing the resilience and stability of the transportation system and indirectly increasing the adjustment capacity of the power system. Therefore, in scenarios involving large-scale SEV scheduling, incentive mechanisms integrating rebalancing incentives have greater practicality.

[0136] (3) Verification of revenue equilibrium under different budget and fleet size conditions To further verify the balance of fleet revenue distribution under different operating conditions, this embodiment analyzes the issue from two dimensions: first, adjusting the budget under a fixed fleet size; and second, proportionally expanding the budget while increasing the fleet size. Revenue distribution is presented from four dimensions: time cost, V2G revenue, rebalancing revenue, and travel preference cost, using radar charts to display the revenue levels of different fleets under these dimensions.

[0137] Figure 9 This chart compares the revenue of shared electric vehicle fleets under different budgets. The first quadrant represents the cost of travel preference deviation, the second quadrant represents the total time cost, the third quadrant represents V2G revenue, and the fourth quadrant represents rebalancing revenue. Figure 9This study demonstrates the revenue distribution of each SEV fleet with a fixed fleet size of 10 and budgets of 20, 25, and 30. It can be observed that as the budget increases from 20 to 30, the overall V2G and rebalancing revenues of each fleet improve, and the bar chart becomes more robust. Simultaneously, the gap between fleets in these two revenues gradually narrows, indicating that with more abundant incentive resources, the system is better able to guide fleets to participate in services in a more balanced manner, resulting in fairer revenue distribution and a more stable game equilibrium. Furthermore, time costs and travel preference costs remain largely consistent, suggesting that these costs are primarily determined by network structure and task allocation, directly related to the budget, and therefore better reflect the stability of the fleet strategies themselves.

[0138] Figure 10 This chart compares the revenue of shared electric vehicle fleets at different scales. The first quadrant represents the cost of travel preference deviation, the second quadrant represents the total time cost, the third quadrant represents V2G revenue, and the fourth quadrant represents rebalancing revenue. Figure 10 The data shows the revenue distribution when the number of fleets is 10, 15, and 20, with corresponding budgets of 20, 30, and 40, maintaining a relatively consistent budget per fleet. As the fleet size increases, the revenue distribution across all dimensions remains stable, especially with no abnormal fluctuations in time cost and travel preference cost, indicating a relatively balanced task allocation among fleets. Despite the increased scale, the distribution of V2G revenue and rebalancing revenue still exhibits good symmetry, with no significant loss of revenue for marginal fleets, demonstrating the system's good resource adjustment capabilities during scaling. The consistent combination structure of the four types of revenue across fleets further confirms the model's good balance and scalability.

[0139] In summary, regardless of different budget levels or in scenarios involving fleet expansion, the distribution of SEV fleet revenue across multiple dimensions maintains good consistency and equilibrium. This indicates that the proposed game theory model and incentive strategy can achieve a balance in revenue distribution under resource constraints and changes in participation scale, demonstrating good scalability.

[0140] Furthermore, to verify the convergence of the proposed algorithm, the analysis begins with the key variables at both the upper and lower layers. Due to the large number of variables in the system, for ease of demonstration and explanation, this embodiment selects Team 1 as a representative and analyzes its convergence performance during the iteration process, specifically including: the personalized incentive price designed by the CSO, the sensitivity response of Team 1, and its final scheduling decision variables.

[0141] (1) Convergence of CSO incentive prices In each outer iteration, the CSO updates the personalized incentive prices it sends to each team based on feedback from the current SEV teams. For Team 1, the incentive is... pd,1 As the iteration progresses, the CSO adjusts the gradient direction accordingly. p d,1 Adjustments were made to more effectively guide the vehicle fleet in V2G discharge and rebalancing operations. Simulation results show that the excitation variable pd,1 reaches stability within a finite number of iterations, indicating that the upper-level optimization objective of the system has good convergence characteristics, such as... Figure 11 As shown.

[0142] (2) Convergence of sensitivity of vehicle No. 1 Sensitivity s This reflects the responsiveness of the SEV fleet to the incentive price offered by the CSO, indicating the changing trend of the SEV fleet's path selection under incentive perturbations. This embodiment uses the sensitivity of the final convergence of fleet number 1. s Taking distribution 1 as an example, a heatmap is used for visualization. The results show that the sensitivity values ​​at most locations approach zero, with only two locations exhibiting weak non-zero values, and these values ​​are small. This indicates that in the convergence state, vehicle 1 no longer responds significantly to the stimulus adjustments at most charging stations and in the rebalancing area, and its behavioral decisions tend to stabilize. Figure 12 As shown, this highly sparse sensitivity structure indicates that after reaching equilibrium, the convoy retains only a weak probability of response to a few policy directions, reflecting the stability and local convergence characteristics of the game solution.

[0143] (3) Convergence of the travel decision of vehicle No. 1 In each iteration, Team 1 selects the optimal charging station, route, and whether to participate in the rebalancing task based on the current incentive price and its own profit maximization principle. Its route decision is determined by variables. x 1 indicates, such as Figure 13 As shown in the figure. Numerical results indicate that the process exhibits a gradual convergence trend during iteration, eventually stabilizing on a set of optimal strategies, demonstrating that the SEV team can respond quickly to stimuli and make stable decisions within the game theory framework.

[0144] Please refer to Figure 2 Embodiment four of the present invention is as follows: A charging load guidance and scheduling system 1 for shared electric vehicles includes a memory 2, a processor 3, and a computer program stored in the memory 2 and executable on the processor 3. When the processor 3 executes the computer program, it implements the various steps of a charging load guidance and scheduling method for shared electric vehicles according to any one of the embodiments 1 to 3.

[0145] In summary, this invention provides a charging load guidance and scheduling method and system for shared electric vehicles. Addressing the incentive design problem of SEV scheduling in an electric-transportation coupled system, it establishes a scheduling model based on the Stackelberg-Nash game, deeply characterizing the game relationship between the system operator (CSO) and the SEV fleet in incentive guidance and strategy response. The model design fully integrates two dimensions: electrical energy incentives and rebalancing incentives. It aims to guide the fleet to make reasonable route selection and rebalancing decisions while ensuring the stable operation of the power system, thereby achieving coordinated optimization of energy supply and demand and traffic flow distribution.

[0146] To address the non-convexity and non-smoothness characteristics of the model, a distributed solution method based on hypergradient descent is proposed. The existence and uniqueness of the Stackelberg-Nash equilibrium are analyzed theoretically to ensure the model's stability. Subsequently, simulation experiments are used to systematically verify the model's performance.

[0147] In simulation experiments, convergence analysis shows that the main decision variables at both the upper and lower levels exhibit good convergence trends during the iteration process, especially demonstrating good stability in key variables such as fleet sensitivity and CSO revenue. In feasibility verification, the differences in traffic congestion and system benefits under different rebalancing incentive strategies were compared and analyzed. Furthermore, the system's adaptability and resource allocation balance under various scenarios were explored, considering different budgets and fleet sizes. Finally, through perturbation experiments on upper and lower level variables, the deviation from the optimal solution was evaluated, verifying that the proposed method can maintain a high level of revenue under different perturbation conditions, demonstrating strong optimality and robustness.

[0148] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent modifications made based on the content of the present invention specification and drawings, or direct or indirect applications in related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for guiding and scheduling charging loads for shared electric vehicles, characterized in that, Including the following steps: Acquire electric transportation coupling system operators and shared electric vehicle fleets; Establish a charging load guidance and scheduling game model based on Stackelberg-Nash game theory between upper-level electric transportation coupling system operators and lower-level shared electric vehicle fleets; The objective function and its constraints of the charging load guidance and scheduling game model are determined. The objective function includes a first objective function that considers the minimum cost of the shared electric vehicle fleet considering charging and discharging incentives and rebalancing incentives, and a second objective function that considers the minimum cost of the electric transportation coupling system operator considering incentive response. The constraints include a first constraint corresponding to the first objective function and a second constraint corresponding to the second objective function. The charging load guidance and scheduling game model is solved in a distributed manner using the super gradient descent algorithm to obtain the charging load guidance and scheduling scheme. The distributed solution using the super gradient descent algorithm includes: the electric transportation coupling system operator obtaining the response and Jacobian matrix of the shared electric vehicle fleet; each shared electric vehicle fleet calculating an auxiliary matrix of projection mapping with respect to its own decision variables, sensitivity, and traffic flow aggregation, updating the sensitivity according to the auxiliary matrix, and feeding back the updated sensitivity to the electric transportation coupling system operator in aggregate form; The first objective function for minimizing the cost of a shared electric vehicle fleet, considering charge / discharge incentives and rebalancing incentives, includes: Considering charging and discharging incentives, calculate the first revenue value of providing electrical energy to a shared electric vehicle fleet; Considering the rebalancing incentives, calculate the second revenue value provided by the shared electric vehicle fleet for rebalancing; Calculate the deviation costs of a shared electric vehicle fleet changing routes or destinations; Calculate the travel time cost of a shared electric vehicle fleet; The sum of the deviation cost and the travel time cost, minus the sum of the first revenue value and the second revenue value, yields the aggregated cost of the shared electric vehicle fleet. The first objective function is obtained by minimizing the aggregation cost of the shared electric vehicle fleet; The calculation of the first revenue value for providing electrical energy to a shared electric vehicle fleet, considering charging and discharging incentives, includes: In the formula, This represents the first profit value. Indicates the location of the charging station. Indicates the electric transportation coupling system operator at the destination d For the team i Excitation of discharge, This indicates the degradation and loss of the battery during the discharge process. Indicates the first i The team selected the first d The probability of a destination. m i Indicates the team i The number of shared electric vehicles owned; The calculation of the second revenue value provided by the shared electric vehicle fleet for rebalancing, considering the rebalancing incentives, includes: In the formula, This represents the second revenue value. Indicates the rebalancing site. Indicates the electric transportation coupling system operator at the destination d For the team i The provided rebalancing incentives This indicates the degradation loss of the battery during the rebalancing process.

2. The charging load guidance and scheduling method for shared electric vehicles according to claim 1, characterized in that, The first constraint includes traffic flow conservation constraint, upper and lower bound constraints on the decision vector of the first objective function, and coupled resource constraint.

3. The charging load guidance and scheduling method for shared electric vehicles according to claim 1, characterized in that, The second objective function for minimizing the cost for operators of electric-transport coupled systems, considering the incentive response, includes: Calculate the total travel time cost for all shared electric vehicle fleets; Calculate the discharge revenue value for the operator of the electric transportation coupling system; Calculate the rebalancing revenue value for the operator of the electric transportation coupling system; The total travel time cost is subtracted from the sum of the discharge revenue and the rebalancing revenue to obtain the operator cost of the electric transportation coupling system. The second objective function is obtained by minimizing the operator cost of the electric transportation coupling system.

4. The charging load guidance and scheduling method for shared electric vehicles according to claim 3, characterized in that, The second constraint includes upper and lower bound constraints on the decision vector of the second objective function and constraints on the total amount of incentive resources.

5. The charging load guidance and scheduling method for shared electric vehicles according to claim 4, characterized in that, The total constraint on incentive resources includes: In the formula, C b Indicates the total amount of incentives. C Indicates the upper limit of the total incentive amount. p It indicates motivation. A vector representing the traffic flow to the destination; Indicates the location of the charging station. This represents the excitation from the electric transportation coupling system operator to the vehicle fleet i at destination d. This indicates the degradation loss of the battery during the discharge process; Indicates the rebalancing site. This represents the rebalancing incentive provided by the electric transportation coupling system operator to fleet i at destination d. This indicates the degradation loss of the battery during the rebalancing process; This indicates the assembly of the convoy. m i Indicates the team i The number of shared electric vehicles owned. Indicates the team i The total flow rate reaching each charging station.

6. The charging load guidance and scheduling method for shared electric vehicles according to claim 1, characterized in that, The charging load guidance and scheduling game model is as follows: In the formula, f i Let Ω represent the first objective function. i This indicates the first constraint condition. Let Ω represent the second objective function. CSO This indicates the second constraint condition. It refers to a fleet of shared electric vehicles, and CSO stands for Electric Transportation Coupling System Operator.

7. The charging load guidance and scheduling method for shared electric vehicles according to claim 6, characterized in that, The game theory of the charging load guidance and scheduling game model includes the following steps: P1. The electric transportation coupling system operator will broadcast the incentive to all shared electric vehicle fleets; P2. The shared electric vehicle fleet determines the optimal response strategy based on the incentives broadcast by the electric transportation coupling system operator, combined with the first objective function and the first constraint conditions. P3. The electric transportation coupling system operator determines the optimal incentive based on the best response strategy of all shared electric vehicle fleets, combined with the second objective function and the second constraint. Repeat steps P1 through P3 until the determined optimal stimulus remains unchanged.

8. A charging load guidance and scheduling system for shared electric vehicles, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements each step of the charging load guidance and scheduling method for a shared electric vehicle as described in any one of claims 1 to 7.

Images