Electric vehicle mobile charging station scheduling method for realizing charge and discharge co-scheduling

Abstract

The invention relates to an electric vehicle mobile charging station scheduling method for realizing charge and discharge collaborative scheduling, which comprises the following steps of: 1, constructing a mobile charging station charging scheduling model based on a multi-agent reinforcement learning algorithm idea; 2, designing a negotiation mechanism between the power grid and the mobile charging station according to reasonable constraint conditions, and constructing a discharging scheduling model of the mobile charging station; 3, integrating the charging and discharging scheduling model, considering the conflict problem of the mobile charging station in response to the charging and discharging request, constructing a charging and discharging collaborative optimization scheduling framework, and scheduling the mobile charging station of the electric vehicle; 4, scheduling the mobile charging station according to the constructed charging and discharging collaborative optimization scheduling framework; according to the scheme, power grid load data is introduced as a constraint condition to promote the mobile charging station to respond to a discharging request, and charging and discharging cooperative scheduling of the mobile charging station is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric vehicle Internet of Things, and in particular relates to a method for scheduling an electric vehicle mobile charging station for realizing coordinated scheduling of charging and discharging. Background Art

[0002] In recent years, as the issue of greenhouse gas emissions has received increasing attention, the market share of electric vehicles has tended to accelerate. Compared with traditional fuel vehicles, electric vehicles are powered by onboard batteries. At present, the main problem that electric vehicles need to solve is still the endurance problem. Therefore, the widespread deployment of fixed charging piles to charge electric vehicles is the main charging strategy today. However, the large-scale deployment of fixed charging piles requires high costs and complex pre-planning, and it is difficult to meet the charging needs of electric vehicles in different scenarios, such as highways and rural areas. Therefore, the concept of mobile charging piles is proposed as a supplementary charging form for fixed charging piles. Compared with fixed charging piles, mobile charging piles are flexible, fast, instant response and low deployment cost. They can not only provide stable charging services, but also adapt to various charging scenarios to meet the charging needs of different regions and time periods. Usually, mobile charging piles will only move to the negotiated position to charge electric vehicles after receiving a charging request. This method will cause problems such as long idle time of mobile charging piles, low charging efficiency, and long waiting time for electric vehicles to serve.

[0003] As the number of electric vehicles grows, their disorderly charging behavior, that is, large-scale charging behavior without scheduling, will add to the peak load and increase the pressure on the power grid. The application of V2G technology can effectively alleviate the pressure on the power grid, that is, electric vehicles are used as mobile distributed energy storage units to discharge back to the power grid. This method can not only reduce peak loads and fill valleys, improve the utilization rate of electric energy, but also increase the benefits of electric vehicle users. However, considering the problems of battery life attenuation, limited benefits, and low level of awareness, today's electric vehicle users are less willing to actively reverse power to the power grid. As a new type of charging equipment, mobile charging piles can not only charge electric vehicles, but also reverse discharge to the power grid during peak load periods, and there is no problem of insufficient subjective willingness. Summary of the invention

[0004] Based on the idea of multi-agent reinforcement learning algorithm, the present invention obtains the charging scheduling model of mobile charging stations by training the location and available power of electric vehicles and mobile charging piles in the urban road network. At the same time, according to reasonable constraints, the negotiation mechanism between the power grid and the mobile charging station is designed to schedule the mobile charging station to respond to the reverse discharge request sent by the power grid, thereby realizing the coordinated scheduling of charging and discharging.

[0005] To achieve the above object, the technical solution of the present invention is as follows: A scheduling method for an electric vehicle mobile charging station that realizes coordinated charging and discharging scheduling, comprising the following steps:

[0006] Step 1: Based on the idea of a multi-agent reinforcement learning algorithm, construct a charging scheduling model for the mobile charging station;

[0007] Step 2: According to reasonable constraint conditions, design a negotiation mechanism between the power grid and the mobile charging station, and construct a discharging scheduling model for the mobile charging station;

[0008] Step 3: Integrate the charging and discharging scheduling models, consider the contradiction problem of the mobile charging station responding to charging and discharging requests, construct a coordinated optimization scheduling framework for charging and discharging, and schedule the electric vehicle mobile charging station;

[0009] Step 4: Schedule the mobile charging station according to the constructed coordinated optimization scheduling framework for charging and discharging.

[0010] As an improvement of the present invention, the charging scheduling model of the mobile charging station based on the idea of a multi-agent reinforcement learning algorithm in Step 1 is as follows:

[0011]

[0012] Let be the reward obtained when the idle mobile charging station ψ j is scheduled to the next position p next at the end of the t-th time slot, and St(v i ) be the power gap generated when the electric vehicle v i travels from the departure position to the target position and the initial power cannot cover the total power required for the electric vehicle to travel; the delay time T c (v i ,ψ j ), T c (v i ,ψ j′ ), where ψ j represents the idle mobile charging station that actively approaches the electric vehicle v i to wait for potential charging opportunities, ψ j′ represents the busy mobile charging station that is performing charging and discharging near the electric vehicle v i , and T c (v i ,ψ j ), T c (v i ,ψ j′ ) is the distance between the electric vehicle v i and the mobile charging station ψ j or ψ j′The delay time required to start charging after reaching the negotiated charging location; Ut(ψ j′ ) (t+1) For the electric vehicle v i Adjacent to the busy mobile charging station ψ j′ The available power that may be used to charge the electric vehicle v i after completing the charge and discharge task, α is a preset weight, Indicates the influence of the electric vehicle v i on the scheduling decision; Indicates the influence of the adjacent busy mobile charging station on the scheduling decision,

[0013]

[0014] Let be the set of all available scheduling positions p of the idle mobile charging station in the road network at the t-th time slot, e(ψ j ) (t) be the remaining power of ψ j at the t-th time slot; c·d(ψ j ,p) is the power required for ψ j to move to the position p. When the remaining power is sufficient, the idle mobile charging station ψ j actively approaches the electric vehicle v i and waits for potential charging opportunities. Some busy mobile charging stations nearby may compete for charging opportunities. The larger, the greater the possibility that the idle mobile charging station ψ j successfully charges v i ; The smaller, the smaller the possibility that the busy mobile charging station ψ i near the electric vehicle v j′ successfully charges v i . Therefore, The larger, the more ψ j should be scheduled to the position p;

[0015] Introduce the federated learning mechanism to clarify the collaborative strategy between agents, improve the global convergence speed through model parameter sharing, and add a federated learning layer to the charging scheduling model:

[0016]

[0017] Among them, θ global is the global model parameter, and θ i is the local model parameter deployed on each mobile charging station. Each mobile charging station regularly uploads the local Q-network parameter θ i to the central controller, and the global parameter θ globalAfter weighted averaging, it is sent to the local. The global parameter aggregation adopts a dynamic weighting strategy based on the local sample size, and the weight coefficient is the training data volume n of each local model i and the total data volume The ratio between

[0018] Therefore, a collaborative reward term is added to the original reward function and modified to:

[0019]

[0020] where λ is the collaborative coefficient; the KL divergence constrains the difference between the global and local policies, making the update direction of all mobile charging pile scheduling policies tend to the global optimum rather than the local optimum. The introduction of the collaborative reward term solves the problems of environmental non-stationarity and individual local optimum traps in multi-agent systems.

[0021] As an improvement of the present invention, the discharge scheduling model of the mobile charging station in step 2 is as follows:

[0022]

[0023] In order to relieve the tight grid load situation and improve the revenue of the mobile charging station scheduling system at the same time, according to reasonable constraint conditions, an effective communication mechanism is adopted to schedule the mobile charging station to discharge. Let be the discharge amount of the idle mobile charging station in response to the reverse discharge request; be the grid load situation at the t-th time slot, Th peak be the grid load tension threshold; e(ψ j ) (t) , are the remaining power of the mobile charging station sending a discharge response to the grid and ending the discharge at the -th time slot respectively; φ d , φ c represent the nearest discharge position to ψ j and the position for charging ψ j itself; Ut(ψ j ) (t) is the available power at the t-th time slot. The mobile charging pile only sends part of the electric energy back to the grid when performing the discharge task,

[0024] When the grid load is tight, broadcast a discharge request:

[0025]

[0026] After receiving the discharge request, the mobile charging station that meets the mobile constraint conditions sends a charging response to the grid, thus reaching a discharge negotiation,

[0027] c·d(ψj , φ d ) is ψ j Travel to φ d The required power, where c is the power required to travel one unit of distance, and the constraint indicates that the remaining power can support ψ j Travel to φ d , and meet the discharge requirement

[0028]

[0029] c·d(φ d , φ c ) is φ d Travel to φ c The required power, and the constraint indicates that the remaining power after performing the discharge task can support ψ j Travel to the nearest position to charge itself:

[0030]

[0031] Among them, in order to prevent the battery life from decaying due to over-discharge, it is necessary to increase the safety redundancy constraint coefficient ξ so that the depth of discharge will not cause the battery life to decay.

[0032] If the grid load tension threshold is fixed and cannot adapt to seasonal or time-of-day changes, predict the future threshold based on historical load data and perform dynamic adjustment:

[0033] Th peak =μ L (t)+k·σ L (t)

[0034] Among them, μ L (t) and σ L (t) are respectively the mean and standard deviation of historical load; k is the dynamic adjustment coefficient. As an improvement of the present invention, the charge-discharge collaborative optimization scheduling framework in step 3 is as follows:

[0035]

[0036] The objective functions of the charge-discharge collaborative optimization scheduling framework are respectively to maximize the total revenue of the system in all time slots, including the charging revenue and discharge revenue of the mobile charging station; and to maximize the utilization rate of the mobile charging station, that is, to maximize the number of busy mobile charging stations, Bs(ψ j ) is the identifier of whether the mobile charging station ψ j is in a busy state. If Bs(ψ j ) = 1, it means that the mobile charging station ψ j is in a busy state; if Bs(ψ j) = 0, the situation is the opposite. There is a conflict problem between the mobile charging station's response to the electric vehicle charging request and the grid discharging request. If the first-come-first-served or revenue-first principle is followed, the urgent needs may be ignored. Therefore, an emergency scoring function can be introduced in the conflict request processing:

[0037]

[0038] Among them, E gap is the power gap of the discharging request; E total is the total demand for the grid's discharging power; T max is the time that the electric vehicle user expects to wait for service; T wait is the time difference between the request sending time slot and the current time slot. High-priority tasks are allowed to interrupt low-priority tasks, and the expected revenue loss of the preempted party is compensated.

[0039] During the peak load period, the weight ω1 is dynamically increased; during the traffic congestion period, the weight ω2 is dynamically increased.

[0040] As an improvement of the present invention, step 4 is specifically as follows:

[0041] The central controller collects the grid load data in real time at the t-th time slot The mobile charging station collects the position p(v i ) (t) and the remaining power e(v i ) (t) data of the electric vehicles within the communication range, as well as the p(ψ j ) (t) and the remaining power e(ψ j ) (t) data of the mobile charging station. The available scheduling positions of the mobile charging station are generated by the trained decentralized charging and discharging collaborative optimization scheduling model. At the same time, during the period of tight grid load, the central controller sends a discharging request to the mobile charging station, and the mobile charging stations that meet the mobile constraint conditions respond to the discharging request.

[0042] An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the scheduling method of the electric vehicle mobile charging station for realizing charging and discharging collaborative scheduling.

[0043] A computer-readable storage medium, on which a computer instruction is stored. When the computer instruction is executed by a processor, it implements the scheduling method of the electric vehicle mobile charging station for realizing charging and discharging collaborative scheduling.

[0044] Compared with the prior art, the present invention takes into account the discharging potential of electric vehicle mobile charging stations. While utilizing the remaining power and location data of electric vehicles and mobile charging, grid load data is introduced as a constraint condition to promote mobile charging stations to respond to discharging requests, realizing coordinated charging and discharging scheduling of mobile charging stations. Description of the Drawings

[0045] Figure 1 It is a flowchart for the scheduling of mobile charging stations;

[0046] Figure 2 It is a schematic structural diagram of the Q network. Detailed Implementation Manner

[0047] Embodiment: The present invention provides a scheduling method for an electric vehicle mobile charging station to achieve coordinated charging and discharging scheduling, including the following steps:

[0048] Step 1: Based on the idea of multi-agent reinforcement learning algorithm, construct a charging scheduling model for mobile charging stations;

[0049] Step 2: According to reasonable constraint conditions, design a negotiation mechanism between the power grid and mobile charging stations, and construct a discharging scheduling model for mobile charging stations;

[0050] Step 3: Integrate the charging and discharging scheduling models, consider the contradiction problem of mobile charging stations responding to charging and discharging requests, construct a coordinated optimization scheduling framework for charging and discharging, and schedule electric vehicle mobile charging stations;

[0051] Step 4: Schedule mobile charging stations according to the constructed coordinated optimization scheduling framework for charging and discharging.

[0052] For the above-mentioned scheduling method for an electric vehicle mobile charging station to achieve coordinated charging and discharging scheduling, Step 1 is specifically as follows:

[0053] The mobile charging station calculates the power shortage St(v i ), delay time T i (v c (v i , ψ j ) and T c (v i , ψ j′ ) as well as the available power Ut(ψ j′ ) of neighboring busy mobile charging stations according to the real-time collected power and location data, and in combination with the road network. Each mobile charging station, as an agent capable of making independent decisions, needs to design the components of reinforcement learning. Its state space includes the remaining power e(ψ (t+1) ). j ) (t) , location p(ψ j ) (t); Its action space includes candidate available scheduling positions p, where and it is necessary to satisfy the movement constraint conditions; its reward function is By quantifying the charging urgency of the electric vehicle v i and by reflecting the competitive impact of neighboring busy mobile charging stations, considering the setting of the reward function structure from the perspective of a potential charging opportunity for an idle mobile charging station ψ j competing with multiple busy mobile charging stations ψ j′ for a potential charging opportunity of an electric vehicle v i The weight coefficient α can adjust the ratio between the two impacts. The Deep Q-Network (DQN) algorithm is used for training to maximize the long-term cumulative reward, and the optimal scheduling position p with the highest score is dynamically selected within each time slot t next , while ensuring that the remaining power of the mobile charging station is sufficient to move to the scheduling position.

[0054] For the aforementioned method for scheduling an electric vehicle mobile charging station to achieve coordinated charging and discharging, step 2 is specifically as follows:

[0055] The grid load monitoring module collects grid load data in real time When the load exceeds the peak threshold Th peak , the grid broadcasts a discharge request. If the mobile charging station that receives the discharge request meets the dual constraints, including the movement feasibility constraint, that is, the remaining power at present (in the t-th time slot) is sufficient to support the mobile charging station to move to the discharge position and meet the discharge demand and the return guarantee, that is, the remaining power after discharging (in the -th time slot) is sufficient to support the mobile charging station to move to the discharge position. The calculation of the discharge amount needs to consider both individual capabilities and global balance, taking the minimum value between the available discharge amount Ut(ψ j ) j ) (t) -c·d(ψ j , φ d )-c·d(φ d , φ c ) of the mobile charging station ψ j and the load regulation amount that ψ needs to bear, and distributing the total discharge demand proportionally to avoid local overload. In addition, a distributed negotiation mechanism is adopted, and the mobile charging station submits the available discharge amount Ut(ψ j ) (t) -c·d(ψ j , φ d )-c·d(φ d , φ cAnd the current location, dynamically allocate the discharge demand task after the comprehensive global remaining power distribution of the power grid.

[0056] For the above-mentioned method for scheduling an electric vehicle mobile charging station to achieve charge-discharge coordinated scheduling, step 3 is specifically as follows:

[0057] By constructing a two-objective optimization framework, integrating the charging and discharging scheduling models, and solving the conflicts of the mobile charging station in responding to charging and discharging requests. Define the objective function: maximize the total system revenue (including charging revenue and discharging revenue) and the utilization rate of the mobile charging station (the number of busy states). The charging revenue is based on the number of electric vehicles served and the electricity price, and the discharging revenue is based on the discharging amount and the grid subsidy; the utilization rate is measured by the number of mobile charging stations marked as busy states. For the conflicts of charging and discharging requests, introduce a dynamic priority rule: compare the expected revenues of charging and discharging, and give priority to responding to the task with higher revenue; if the power shortage of the electric vehicle exceeds the threshold, forcefully interrupt the low-priority discharging task to ensure the emergency charging demand. In addition, adopt a decentralized multi-agent collaborative decision-making framework. Each mobile charging station acts as an independent agent, and based on local information (power, location) and global goals (total revenue, utilization rate), perform dynamic scheduling. The central controller coordinates the task allocation through an auction mechanism (for example, the charging task is preferentially allocated to the charging station that is close and has high revenue, and the discharging task is preferentially allocated to the charging station with sufficient power and low moving cost), so as to achieve efficient resource utilization and conflict coordination.

[0058] For the above-mentioned method for scheduling an electric vehicle mobile charging station to achieve charge-discharge coordinated scheduling, step 4 is specifically as follows:

[0059] Execute charge-discharge coordinated scheduling through real-time data collection, dynamic task allocation, and reinforcement learning feedback. The central controller monitors the grid load data in real time. If it exceeds the peak threshold, it broadcasts a discharging request and simultaneously receives the location, remaining power of the mobile charging station, and the power shortage information of the electric vehicle. When a charging task is generated, calculate the optimal moving position of the idle charging station based on the scheduling model in step 1 to ensure that its remaining power meets the moving requirements; when a discharging task is triggered, the mobile charging station needs to verify the double constraints (the current power is sufficient to move to the discharging position and complete the discharging, and the remaining power after discharging is sufficient to return). The charging stations that meet the conditions submit the available discharging amount, and the grid dynamically allocates tasks proportionally. At the same time, use the DQN algorithm to continuously optimize the scheduling strategy, update the reinforcement learning model according to the historical scheduling data, and improve the long-term cumulative reward. In case of abnormal situations (such as insufficient power or path blockage), the central controller reallocates tasks and adjusts the subsequent strategy.

[0060] Figure 1It is a scheduling flowchart of a mobile charging station. In the t-th time slot, the local controller deployed in the mobile charging station collects the remaining power and real-time positions of neighboring electric vehicles and busy mobile charging stations for scheduling the mobile charging station to perform charging tasks. The central controller collects grid load data for initiating the broadcast of grid discharge requests. Mobile charging stations that meet certain constraint conditions respond to charge and discharge requests. If charge and discharge requests are received simultaneously in the t-th time slot, service conflicts are avoided by comparing the benefits.

[0061] It should be noted that the above embodiments are not intended to limit the protection scope of the present invention. Equivalent transformations or substitutions made on the basis of the above technical solutions all fall within the protection scope of the claims of the present invention.

Claims

1. A scheduling method for an electric vehicle mobile charging station that realizes coordinated charging and discharging scheduling, characterized in that, It includes the following steps: Step 1: Based on the idea of multi-agent reinforcement learning algorithm, construct a charging scheduling model for mobile charging stations; Step 2: According to reasonable constraint conditions, design a negotiation mechanism between the power grid and mobile charging stations, and construct a discharging scheduling model for mobile charging stations; Step 3: Integrate the charging and discharging scheduling models, consider the conflict problem of mobile charging stations responding to charging and discharging requests, construct a charging and discharging collaborative optimization scheduling framework, and schedule electric vehicle mobile charging stations; Step 4: Schedule the mobile charging stations according to the constructed charging and discharging collaborative optimization scheduling framework.

2. The method for scheduling an electric vehicle mobile charging station to achieve coordinated charging and discharging according to claim 1, characterized in that, The charging scheduling model of the mobile charging station based on the idea of multi-agent reinforcement learning algorithm in Step 1 is as follows: set up is the number of idle mobile charging stations ψ at the end of the tth time slot j Dispatched to the next position p next The reward obtained, St(v i ) is an electric vehicle v i When the initial power cannot cover the total power required by the electric vehicle from the starting point to the target point, the power gap is generated; the delay time T c (v i ,ψ j ), T c (v i ,ψ j′ ), where ψ j Indicates active approach to electric vehicle v i An idle mobile charging station waiting for a potential charging opportunity, ψ j′ Indicates electric vehicle v i A busy mobile charging station is charging and discharging nearby. c (v i ,ψ j ), T c (v i ,ψ j′ ) is an electric vehicle v i With mobile charging stationψ j or j′ After the location negotiation, the delay time required to reach the negotiated charging location to start charging; Ut(ψ j′ ) (t+1) For electric vehicles i Near busy mobile charging stationsψ j′ After completing the charging and discharging tasks, it may be used to power electric vehicles i The available power for charging, α is the preset weight, Indicates electric vehicle v i Impact on scheduling decisions; represents the impact of nearby busy mobile charging stations on the scheduling decision, Let be the set of all available scheduling positions p of the idle mobile charging station in the road network at the t-th time slot, e(ψ j ) (t) be the remaining power of ψ j at the t-th time slot; c·d(ψ j , p) be the power required for ψ j to move to position p. When the remaining power is sufficient, the idle mobile charging station ψ j actively approaches the electric vehicle v i to wait for potential charging opportunities. Some nearby busy mobile charging stations may compete for charging opportunities. The larger it is, the greater the possibility that the idle mobile charging station ψ j successfully charges v i ; The smaller it is, the smaller the possibility that the busy mobile charging station ψ i near the electric vehicle v j′ successfully charges v i . Therefore, the larger it is, the more ψ j should be scheduled to position p; Introduce a federated learning mechanism to clarify the collaborative strategy between agents, improve the global convergence speed through model parameter sharing, and add a federated learning layer to the charging scheduling model: where, θ global is the global model parameter, and θ i is the local model parameter deployed on each mobile charging station. Each mobile charging station regularly uploads the local Q-network parameter θ i to the central controller. After global parameter θ global is weighted and averaged, it is sent to the local side. The global parameter aggregation adopts a dynamic weighting strategy based on the local sample size, and the weight coefficient is the ratio between the training data volume n i of each local model and the total data volume ​ Therefore, add a collaborative reward term to the original reward function and modify it to: where λ is the collaborative coefficient; the KL divergence constrains the difference between the global and local policies, making the update directions of all mobile charging pile scheduling policies tend to the global optimum rather than the local optimum. The introduction of the collaborative reward term solves the problems of environmental non-stationarity and individual local optimum traps in multi-agent systems.

3. The electric vehicle mobile charging station scheduling method for realizing coordinated charging and discharging scheduling according to claim 1, characterized in that, The discharging scheduling model of the mobile charging station in Step 2 is as follows: To alleviate the tight grid load situation and improve the revenue of the mobile charging station scheduling system, according to reasonable constraints, an effective communication mechanism is adopted to schedule the mobile charging station to discharge. Let be the discharge amount of the idle mobile charging station in response to the reverse discharge request; be the grid load situation at the t-th time slot, Th peak be the grid load tension threshold; e(ψ j ) (t) and are respectively the remaining power of the mobile charging station when sending a discharge response to the grid at the t-th time slot and when ending the discharge at the -th time slot; φ d and φ c represent the nearest discharge position to ψ j and the position for charging ψ j itself; Ut(ψ j ) (t) is the available power at the t-th time slot. The mobile charging pile only delivers part of the electric energy to the grid when performing the discharge task. When the power grid load is tight, broadcast a discharging request: The mobile charging station receives the discharging request, and the mobile charging stations that meet the mobile constraint conditions send a charging response to the power grid, thus achieving discharging negotiation. c·d(ψ j , φ d ) is the electric energy required for ψ j to travel to φ d , where c is the electric energy required to travel one unit of distance, and the constraint condition indicates that the remaining electric energy can support ψ j to travel to φ d , and meet the discharge amount requirement c·d(φ d , φ c ) is φ d The electric energy required to travel to φ c The remaining electric energy after performing the discharging task can support ψ as indicated by the constraint condition j Travel to the nearest position to charge itself: Among them, in order to prevent battery life attenuation caused by over-discharging, it is necessary to increase the safety redundancy constraint coefficient ξ so that the discharging depth will not cause battery life attenuation. If the power grid load tight threshold is fixed and cannot adapt to seasonal or time-of-day changes, predict the future threshold based on historical load data and make dynamic adjustments: Th peak = μ L (t) + k·σ L (t) Among them, μ L (t) and σ L (t) are the mean and standard deviation of the historical load respectively; k is the dynamic adjustment coefficient.

4. The method for scheduling an electric vehicle mobile charging station for coordinated charging and discharging according to claim 1, characterized in that The charging and discharging collaborative optimization scheduling framework in Step 3 is as follows: The objective functions of the charge-discharge collaborative optimization scheduling framework are respectively to maximize the total revenue of the system in all time slots, including the charging revenue and discharging revenue of mobile charging stations; and to maximize the utilization rate of mobile charging stations, that is, to maximize the number of mobile charging stations in the busy state, Bs(ψ j ) is the identifier of whether the mobile charging station ψ j is in the busy state. If Bs(ψ j ) = 1, it means that the mobile charging station ψ j is in the busy state; if Bs(ψ j ) = 0, the situation is opposite. There is a conflict problem between the mobile charging station's response to the electric vehicle charging request and the power grid discharging request. If the first-come-first-served or revenue-first principle is followed, the emergency demand is ignored. Therefore, an emergency scoring function is introduced in the conflict request handling: Among them, E gap is the power gap of the discharge request; E total is the total demand for the discharge amount of the power grid; T max is the time that electric vehicle users are expected to wait for service; T wait is the time difference between the request sending time slot and the current time slot, allowing high-priority tasks to interrupt low-priority tasks and compensating for the expected revenue loss of the preemptor. During peak load periods, dynamically increase the weight ω1; during traffic congestion periods, dynamically increase the weight ω2, and update the weights in real time through online learning to adapt to the power grid load fluctuations and the changes in electric vehicle user demands.

5. The method for scheduling an electric vehicle mobile charging station to achieve coordinated charging and discharging according to claim 1, wherein Step 4 is specifically as follows: The central controller collects the grid load data in real time at the t-th time slot The mobile charging station collects the position p(v of the electric vehicle in real time within the communication range i ) (t) and the remaining battery level e(v i ) (t) data, as well as the p(ψ of the mobile charging station j ) (t) and the remaining battery level e(ψ j ) (t) data. The available scheduling positions of the mobile charging stations are generated by the trained decentralized charge-discharge collaborative optimization scheduling model. At the same time, during the period of tight grid load, the central controller sends a discharge request to the mobile charging stations, and the mobile charging stations that meet the mobile constraint conditions respond to the discharge request.

6. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, When the processor executes the program, it implements the scheduling method for the electric vehicle mobile charging station to achieve charging and discharging collaborative scheduling as described in any one of claims 1 to 5 above.

7. A computer-readable storage medium having computer instructions stored thereon, characterized in that, When the computer instruction is executed by the processor, it implements the scheduling method for the electric vehicle mobile charging station to achieve charging and discharging collaborative scheduling as described in any one of claims 1 - 5.

Images