Electric vehicle intelligent scheduling method and system based on dynamic charging

Abstract

The invention provides an electric vehicle intelligent scheduling method and system based on deep reinforcement learning, and the method comprises the steps: collecting power grid information and electric bus information, and carrying out the preprocessing of the power grid information and the electric bus information; constructing an electric vehicle intelligent scheduling model, wherein the electric vehicle intelligent scheduling model is obtained after training and updating based on a deep reinforcement learning model; based on the power grid information, the electric bus information and the electric bus intelligent scheduling model, executing the scheduling process of the electric bus; based on a scheduling result, evaluating and updating the intelligent scheduling model of the electric vehicle; according to factors such as the power grid load condition and the passenger queuing condition, intelligent planning of the charging amount, the charging mode and the like of the electric bus is achieved, the electric bus can make the maximum charging selection, and the charging cost and the passenger queuing duration are minimized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of electric vehicle management, and particularly relates to a charging scheduling method and system for electric vehicles, a storage medium and an electronic device. BACKGROUND

[0002] With the continuous development of new energy technology and the acceleration of urbanization process, new energy vehicles represented by electric vehicles are becoming the main choice for people to purchase and travel. Among them, electric buses as a kind of green and environmentally friendly public transport have been widely used in many areas. At the same time, the large-scale application of electric buses also brings new challenges to power grid management and existing charging scheduling methods. The existing charging management method of electric buses often chooses to charge the public electric buses to full power at the charging station before driving, without considering the maximization of the social benefits of public electric buses, which may cause long charging waiting time, long passenger waiting time and other problems, and does not fully consider the charging cost, power grid load balancing and other factors. SUMMARY

[0003] The present application aims to provide an intelligent scheduling method and system for electric vehicles based on dynamic charging, which can reduce the charging waiting time of electric buses and the passenger waiting time while balancing the power grid load.

[0004] The present application provides an intelligent scheduling method for electric vehicles based on dynamic charging, comprising the following steps: Step one, collecting power grid information and electric bus information, and preprocessing the power grid information and the electric bus information.

[0005] Specifically, the power grid information includes power grid load information and peak-valley electricity price information, and the electric bus information includes electric quantity information, location information, road condition information, passenger queue situation, passenger flow of bus station and other information related to electric bus operation of each electric bus.

[0006] Further, the electric quantity information and location information of electric buses can be obtained through sensors provided by the vehicles, the passenger queue situation and passenger flow of bus station can be obtained through passenger flow monitoring equipment of bus station, and the road condition information can be obtained based on the vehicle-road cooperation system.

[0007] Further, the power grid information and the electric bus information can also be subjected to preprocessing operations such as cleaning, filtering and normalization.

[0008] Step two, constructing an intelligent scheduling model for electric vehicles, which is obtained based on deep reinforcement learning model training and updating.

[0009] Specifically, the intelligent scheduling model of electric vehicles is constructed, including initializing the deep reinforcement learning model: Define the state space S: including the current time period, the electricity price of the current time period, the state of charge of each electric bus, the available charging time, and the passenger queue index. The available charging time is the remaining time from the latest charging end time. The passenger queue index is the passenger queue index of each electric bus calculated by the preset passenger queue index model according to the passenger queue situation, the passenger flow of the bus stop, and the arrival time of the electric bus. The passenger queue index reflects the influence of the passenger queue situation on the charging strategy.

[0010] Define the action space A: including the charging selection of each electric bus in the current time period, including selecting the charging amount, and / or selecting the charging mode, including high-power fast charging and ordinary charging. Define the reward function R: comprehensively consider the charging cost, the grid load balancing, and the passenger queue index. Further, the intelligent scheduling model of electric vehicles is constructed, which further includes policy updating and training of the deep reinforcement learning model: The agent selects an action a from the action space A according to the current state s, executes the action, observes the reward r feedback from the environment, and moves to a new state s'. The process is iterated until the agent can select the action that maximizes its reward. The process of feedback reward is Markov Decision Process (MDP). Step three, based on the grid information, electric bus information, and intelligent scheduling model of electric vehicles, the scheduling process of electric buses is executed.

[0011] Specifically, the scheduling process includes: obtaining the state information of the current time, including the electricity price of the current time period, the state of charge of each electric bus, the available charging time, and the passenger queue index; selecting the optimal action according to the current state information, i.e. determining the charging selection of each electric bus in the current time period; controlling the electric bus to perform the corresponding charging operation according to the selected optimal action, to realize the optimization of charging scheduling.

[0012] Step four, based on the scheduling result, the intelligent scheduling model of electric vehicles is evaluated and updated.

[0013] Specifically, after the charging scheduling is completed, the grid load fluctuation is obtained, and the total charging cost and the passenger queue index are calculated to evaluate the performance of the intelligent scheduling model. According to the evaluation result, the parameters of the intelligent scheduling model are adjusted and optimized to further improve the performance and effect of the system.

[0014] Corresponding to the intelligent scheduling method for electric vehicles based on dynamic charging, the present invention also provides an intelligent scheduling system for electric vehicles based on dynamic charging, comprising: The information acquisition module is used to collect power grid information and electric bus information, and to preprocess the power grid information and electric bus information.

[0015] The model building module is used to build an intelligent scheduling model for electric vehicles, which is obtained by training and updating a deep reinforcement learning model.

[0016] The scheduling module is used to execute the scheduling process for electric buses based on grid information, electric bus information, and an intelligent scheduling model for electric vehicles.

[0017] The evaluation and update module is used to evaluate and update the intelligent scheduling model for electric vehicles based on the scheduling results.

[0018] The present invention also provides an electronic device comprising: a memory and a processor, the memory and the processor being coupled; the memory storing program instructions, which, when executed by the processor, cause the electronic device to perform the intelligent scheduling method for electric vehicles based on dynamic charging of the present invention.

[0019] The present invention also provides a computer-readable storage medium including a computer program that, when run on an electronic device, causes the electronic device to execute the intelligent scheduling method for electric vehicles based on dynamic charging of the present invention.

[0020] This invention provides an intelligent scheduling method and system for electric vehicles based on deep reinforcement learning. The method involves collecting grid information and electric bus information, preprocessing the grid information and electric bus information, constructing an intelligent scheduling model for electric vehicles (EVs) based on deep reinforcement learning training and updates, executing the scheduling process for electric buses based on the grid information, electric bus information, and the intelligent scheduling model, evaluating and updating the intelligent scheduling model based on the scheduling results, and intelligently planning the charging amount and charging mode of electric buses according to factors such as grid load and passenger queuing time. This enables electric buses to make optimal charging choices, minimizing charging costs and passenger queuing time. Attached Figure Description

[0021] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0022] Figure 1 This is a schematic diagram of the intelligent scheduling method for electric vehicles based on dynamic charging according to the present invention.

[0023] Figure 2 This is a schematic diagram of the intelligent scheduling system for electric vehicles based on dynamic charging, as described in this invention. Detailed Implementation

[0024] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0025] Existing electric bus charging management methods often result in problems such as excessively long charging wait times and long passenger waiting times, and do not fully consider factors such as charging costs and grid load balancing.

[0026] Based on this, the purpose of the present invention is to provide an intelligent scheduling method and system for electric vehicles based on dynamic charging, so as to reduce the charging waiting time and passenger waiting time of electric buses, while balancing the grid load.

[0027] This invention provides an intelligent scheduling method for electric vehicles based on dynamic charging, comprising the following steps: Step 1: Collect power grid information and electric bus information, and preprocess the power grid information and electric bus information.

[0028] Specifically, the power grid information includes power grid load information and peak-valley electricity price information, and the electric bus information includes information related to the operation of electric buses, such as power consumption information, location information, road condition information, passenger queuing status, and passenger flow at bus stops.

[0029] Furthermore, information such as the electric bus's battery level and location can be obtained through the vehicle's onboard sensors; passenger queuing status and passenger flow at bus stops can be obtained through passenger flow monitoring equipment at bus stops; and road condition information can be obtained based on a vehicle-to-infrastructure (V2I) system. The power grid information and the electric bus information are transmitted to the scheduling control system in real time or periodically.

[0030] Furthermore, preprocessing operations such as cleaning, filtering, and normalization can be performed on the power grid information and the electric bus information to adapt them to subsequent processing steps and improve information quality and usability. Step 2: Construct an intelligent scheduling model for electric vehicles, which is obtained by training and updating a deep reinforcement learning model.

[0031] Specifically, constructing an intelligent scheduling model for electric vehicles includes initializing a deep reinforcement learning model: Define a state space S, including the current time period, the electricity price during the current time period, the battery status of each electric bus, the available charging time, and the passenger queuing index. The available charging time is the remaining time before the latest charging deadline. The passenger queuing index is a pre-defined passenger queuing index model that calculates the passenger queuing index for each electric bus based on passenger queuing conditions, passenger flow at bus stops, and the arrival time of electric buses. The passenger queuing index reflects the degree to which passenger queuing conditions affect the charging strategy.

[0032] Define action space A: including the charging options for each electric bus during the current time period, including selecting the amount of charge and / or selecting the charging mode, including high-power fast charging and normal charging; Define the reward function R: taking into account charging costs, grid load balancing, and passenger queuing index; the specific expression of the reward function is: R = w1 × R_cost + w2 × R_balance + w3 × R_queue Where w1, w2, and w3 are weighting coefficients, representing the relative importance of charging cost, grid load balance, and passenger queuing index in the reward function, respectively; R_cost is the charging cost-related reward, which is 0 when the electric bus is not charging (i.e., the charging amount is selected as 0), and -Charge × C when charging (where Charge is the charging amount, ...). , Power consumption during normal charging. The additional battery loss calculated using a preset loss model during high-power fast charging (where C is the current electricity price) is calculated. If charging is completed within the specified time, K (K is a positive integer) is added; otherwise, M (M is a positive integer) is subtracted. R_balance is a reward related to grid load balancing. When the grid load is too high, electric buses are encouraged not to charge or choose regular charging, and a corresponding reward is given. R_queue is a reward related to passenger queuing index. The higher the passenger queuing index, the more likely passengers are to choose not to charge or complete charging as quickly as possible (reducing the charging amount and / or choosing high-power fast charging) to reduce passenger waiting time, and a corresponding reward is given. By providing rewards based on charging costs, grid load balancing, and passenger waiting time, electric buses can balance charging choices and passenger waiting situations, maximizing the social benefits of electric buses.

[0033] Furthermore, building an intelligent scheduling model for electric vehicles also includes updating and training the deep reinforcement learning model according to the policy: The agent selects action a from action space A based on the current state s, executes the action, observes the reward r from the environment, and transitions to a new state s'. This process is iteratively executed until the agent can select the action that maximizes its reward. The reward feedback process is a Markov Decision Process. The reinforcement learning model is trained using a Deep Q-Network (DQN) algorithm, and training stability is ensured through a priority experience replay method and a target network. Specifically, experience replay stores a quadruple of historical states, actions, rewards, and new states, and randomly selects mini-batch samples for network updates to avoid training instability caused by excessive correlation. The target network is updated periodically to generate stable target Q-values, reducing the correlation between the current Q-value and the target Q-value, thereby improving training efficiency and stability.

[0034] Step 3: Based on grid information, electric bus information, and the electric vehicle intelligent scheduling model, execute the scheduling process for electric buses.

[0035] Specifically, the scheduling process includes: obtaining the current status information, including the current electricity price, the battery status of each electric bus, the available charging time, and the passenger queuing index; selecting the optimal action based on the current status information, that is, determining the charging option for each electric bus in the current time period; and controlling the electric buses to perform the corresponding charging operation based on the selected optimal action, thereby optimizing the charging schedule.

[0036] Step four: Based on the scheduling results, evaluate and update the intelligent scheduling model for electric vehicles.

[0037] Specifically, after the charging schedule is completed, the grid load fluctuation is acquired, and the total charging cost and passenger average queuing index are calculated to evaluate the performance of the intelligent scheduling model. Based on the evaluation results, the parameters of the intelligent scheduling model are adjusted and optimized to further improve the system's performance and effectiveness. Further adjustments and optimizations to the parameters of the intelligent scheduling model include: adjusting the weight coefficients w1, w2, and w3 in the reward function, and / or optimizing the passenger queuing index model and the loss model.

[0038] Corresponding to the intelligent scheduling method for electric vehicles based on dynamic charging, the present invention also provides an intelligent scheduling system for electric vehicles based on dynamic charging, comprising: The information acquisition module is used to collect power grid information and electric bus information, and to preprocess the power grid information and electric bus information.

[0039] The model building module is used to build an intelligent scheduling model for electric vehicles, which is obtained by training and updating a deep reinforcement learning model.

[0040] The scheduling module is used to execute the scheduling process for electric buses based on grid information, electric bus information, and an intelligent scheduling model for electric vehicles.

[0041] The evaluation and update module is used to evaluate and update the intelligent scheduling model for electric vehicles based on the scheduling results.

[0042] The present invention also provides an electronic device comprising: a memory and a processor, the memory and the processor being coupled; the memory storing program instructions, which, when executed by the processor, cause the electronic device to perform the intelligent scheduling method for electric vehicles based on dynamic charging of the present invention.

[0043] The present invention also provides a computer-readable storage medium including a computer program that, when run on an electronic device, causes the electronic device to execute the intelligent scheduling method for electric vehicles based on dynamic charging of the present invention.

[0044] In summary, this invention provides an intelligent scheduling method and system for electric vehicles based on deep reinforcement learning. It collects grid information and electric bus information, preprocesses the grid information and electric bus information, constructs an intelligent scheduling model for electric vehicles (EVs), which is obtained after training and updating based on a deep reinforcement learning model, executes the scheduling process for electric buses based on the grid information, electric bus information, and the intelligent scheduling model, evaluates and updates the intelligent scheduling model based on the scheduling results, and achieves intelligent planning of charging amount and charging mode for electric buses according to factors such as grid load and passenger queuing time. This enables electric buses to make optimal charging choices, minimizing charging costs and passenger queuing time.

[0045] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the specific details described above.

Claims

1. A dynamic charging based electric vehicle intelligent scheduling method, characterized in that, The method comprises the following steps: Step 1: collecting power grid information and electric bus information, and preprocessing the power grid information and the electric bus information; Step 2: constructing an electric vehicle intelligent scheduling model based on a deep reinforcement learning model training and updating; Step 3: based on the power grid information, the electric bus information and the electric vehicle intelligent scheduling model, performing the scheduling process of the electric bus; Step 4: based on the scheduling result, evaluating and updating the electric vehicle intelligent scheduling model.

2. The method of claim 1, wherein, The power grid information comprises power grid load information and peak-valley electricity price information, and the electric bus information comprises electric quantity information, position information, road condition information, passenger queue condition and passenger flow of each electric bus.

3. The method of claim 1, wherein, The step 2 comprises the following steps: defining a state space S, which comprises a current time period, an electricity price of the current time period, an electric quantity state of each electric bus, a charging available time and a passenger queue index; the charging available time is a remaining time from a latest charging end time; defining an action space A, which comprises a charging selection of each electric bus in the current time period, including selecting a charging quantity and / or selecting a charging mode, including high-power fast charging and ordinary charging; defining a reward function R, which comprehensively considers charging cost, power grid load balancing and the passenger queue index; an agent selects an action a from the action space A according to a current state s, and performs the action, observes a reward r fed back by the environment, and moves to a new state s', and iteratively performs the process until the agent can select an action that maximizes its reward.

4. The method of claim 3, wherein, The step 3 comprises the following steps: obtaining state information of a current time, including an electricity price of a current time period, an electric quantity state of each electric bus, a charging available time and a passenger queue index; selecting an optimal action according to the current state information, and determining a charging strategy of each electric bus in the current time period; controlling the electric bus to perform corresponding charging operation according to the selected optimal action.

5. The method of claim 1, wherein, The step 4 comprises the following steps: after the charging scheduling ends, obtaining power grid load fluctuation, and calculating total charging cost and a passenger queue index, to evaluate the performance of the intelligent scheduling model; and adjusting and optimizing parameters of the intelligent scheduling model according to the evaluation result.

6. The method of claim 3, wherein: the passenger queue index is a passenger queue index calculated by a preset passenger queue index model according to the passenger queue condition, the passenger flow of the bus station, and the arrival time of the electric bus.

7. The method of claim 3, wherein: the feedback reward process is a Markov decision process.

8. A dynamic charging based electric vehicle intelligent scheduling system, characterized in that, The method comprises the following steps: an information collection module, configured to collect power grid information and electric bus information, and preprocess the power grid information and the electric bus information; a model construction module, configured to construct an electric vehicle intelligent scheduling model based on a deep reinforcement learning model training and updating; and a model evaluation module, configured to evaluate and update the electric vehicle intelligent scheduling model based on a scheduling result. The scheduling module is configured to perform a scheduling process of the electric bus based on power grid information, electric bus information and an electric vehicle intelligent scheduling model. The evaluation updating module is configured to evaluate and update the electric vehicle intelligent scheduling model based on a scheduling result.

9. An electronic device, comprising: The electronic device comprises a memory and a processor, which are coupled; the memory stores program instructions, and the program instructions are executed by the processor to enable the electronic device to perform the dynamic charging based electric vehicle intelligent scheduling method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer program enables an electronic device to perform the dynamic charging based electric vehicle intelligent scheduling method according to any one of claims 1 to 7 when the computer program is run on the electronic device. The computer program enables an electronic device to perform the dynamic charging based electric vehicle intelligent scheduling method according to any one of claims 1 to 7 when the computer program is run on the electronic device.

Images