Automatic parking charging system

Through sensor network and deep reinforcement learning algorithms, real-time path planning and charging power adjustment of automatic parking systems in complex environments is realized, solving the problems of parking lot environment changes and battery health status adjustments, and improving charging efficiency and battery life.

CN118906859AActive Publication Date: 2024-11-08SINOHYDRO POWER (BEIJING) CO LTD

Patent Information

Application Number
CN202411250348.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-06
Publication Date
2024-11-08
Estimated Expiration
2044-09-06

AI Technical Summary

Technical Problem

The existing automatic parking charging system cannot adapt to changes in the parking lot environment in real time, resulting in the vehicle being unable to quickly align the charging coils, and the charging power cannot be personalized according to the battery health status, affecting charging efficiency and battery life.

Method used

The sensor network is used to collect environmental data in real time, combine the deep reinforcement learning algorithm to dynamically plan the path, and obtain the vehicle battery status through the battery management system, adjust the charging power, and ensure accurate alignment between the vehicle and the charging coil and safe charging.

Benefits of technology

It improves charging efficiency and system adaptability, extends battery life, reduces the need for repeated adjustments of vehicles, and improves charging safety and overall efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118906859A_ABST
    Figure CN118906859A_ABST
Patent Text Reader

Abstract

The invention discloses an automatic parking charging system, and relates to the technical field of automatic parking, the optimal parking position of vehicles of the same model is effectively predicted through historical charging data analysis, the relative position of the vehicle is monitored in real time, the vehicle is accurately aligned with a charging coil, even if slight displacement occurs, the system can automatically perform fine adjustment, and the charging efficiency is improved. Manual repeated adjustment is avoided, the charging efficiency is greatly improved, meanwhile, deep reinforcement learning is introduced, the vehicle can carry out path planning in a real-time dynamic environment, the parking route is flexibly adjusted, interference of other vehicles or obstacles is avoided, the efficiency of the system in a complex scene is improved through the self-adaptive capacity, the requirement for repeatedly adjusting the path of the vehicle is reduced, and the system reliability is improved. The battery health state is monitored in real time through the battery management system BMS, the charging power is dynamically adjusted according to the temperature and the charging cycle index data, the service life of the battery is prolonged, overheating is prevented, and the overall charging safety is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of automatic parking, in particular to an automatic parking charging system. Background Art

[0002] As an important technology for the development of electric vehicles, the automatic parking and charging system aims to solve the problems of difficult parking and complicated charging through automatic parking and wireless charging.

[0003] The current automatic parking charging system requires the vehicle to be parked precisely above the wireless charging coil to achieve the highest charging efficiency. However, the current system uses a fixed path planning algorithm based on sensors (such as cameras, ultrasonic sensors, etc.), such as A or Dijkstra algorithm, to guide the vehicle to park automatically. However, the sensor is greatly affected by changes in the parking lot environment, such as obstacles and the movement of other vehicles. It is difficult to adjust the parking path in real time, resulting in the vehicle being unable to fully align with the charging coil, affecting the charging efficiency. The parking path cannot be adjusted quickly and often needs to be replanned.

[0004] At present, the optimization method for path planning generally adopts the method of increasing sensor accuracy to optimize the parking path. Even if the sensor accuracy is improved, it is still difficult to quickly respond to the dynamic behavior of other vehicles or pedestrians in a complex parking environment. In order to align the charging coil, the parking space needs to be adjusted repeatedly, which wastes time and reduces customer satisfaction. In addition, during the charging process of parking, the battery health status of the electric vehicle will also affect the charging efficiency. The existing automatic charging system uses a unified charging power and cannot perform personalized charging power allocation according to the battery health status of different vehicles. The aging of the vehicle battery will affect the charging rate and safety. If the battery condition is not good but high power charging is still used, it will accelerate battery aging and even pose a safety risk. Therefore, an automatic parking charging system is urgently needed to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides an automatic parking charging system to solve the problems of being unable to adapt to changes in paths and parking spaces in the parking lot in real time, being unable to quickly align the charging coil, resulting in insufficient parking accuracy and efficiency, and being unable to provide personalized adjustment of charging power in charging distribution, and being unable to protect battery health and maximize charging efficiency.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In one aspect, the present invention provides an automatic parking charging system, comprising:

[0009] A sensor network, which is used to collect environmental data in the parking lot in real time. The network consists of lidar, cameras, and ultrasonic radar;

[0010] A battery management BMS system data acquisition unit, which acquires the vehicle battery status through the vehicle's BMS system;

[0011] A path planning module that adjusts the parking path and aligns the vehicle with the charging coil in real time based on lidar and camera data;

[0012] The charging dynamic fine-tuning module is used to fine-tune the relative position of the vehicle and the charging coil during the charging phase. If the vehicle is slightly displaced, the vehicle position is adjusted to keep it aligned with the charging coil.

[0013] Charging power regulation module, which adjusts the charging power according to the battery status of the vehicle

[0014] The intelligent outbound module is used to move the charged vehicles to the outbound area.

[0015] In another aspect, the present invention provides an automatic parking charging method, comprising:

[0016] Step S1, data collection inside the parking lot,

[0017] A sensor network is deployed in the parking lot to collect environmental data in real time, and obtain the vehicle battery status through the vehicle battery management system BMS;

[0018] Step S2: Parking path confirmation.

[0019] Based on environmental data, a dynamic path planning algorithm based on deep reinforcement learning is used to adjust the parking path. The environmental data is real-time environmental data, including path blockages caused by the movement of other vehicles;

[0020] At the same time, the historical charging relative position behavior pattern analysis of vehicles of the same model is carried out to predict the best parking spot for the vehicle and align the charging coil of the vehicle with the parking space;

[0021] Step S3, charging power distribution,

[0022] After the vehicle is parked according to the path determined in step S2, the battery health status data obtained from the battery management system BMS is analyzed, and the charging mode and power of the charging pile are adjusted according to the battery status; the battery is prevented from aging or overheating during charging, and a lower power is automatically allocated to the battery in poor health status to ensure charging safety;

[0023] Step S4: Parking space management after charging is completed.

[0024] After charging is completed, the vehicle's stay time is managed according to the parking lot demand and traffic flow. If there are other vehicles waiting to be charged, the vehicles that have completed charging will be arranged to move to the outbound area.

[0025] Furthermore, environmental data includes vehicle location and movement, parking space occupancy, charging station status, and path obstruction;

[0026] The vehicle battery status includes battery temperature, remaining power, and number of charging cycles.

[0027] Furthermore, the parking path adjustment method in step S2 is:

[0028] The parking environment is modeled as a state space S, where each state s t ∈S, at time step t, in, represents the current position coordinates of the vehicle at time t, Indicates the vehicle's heading angle. represents the obstacle coordinates, O t represents obstacles in the parking lot, P t represents the position coordinates of the target parking space, P t It is fixed during parking;

[0029] Define the vehicle's action space A, which is the set of actions performed by the vehicle at each time step, a t ∈A,a t =(Δv t ,Δθ t ), where Δv t represents the change in vehicle speed at time t, Δθ t represents the change in the steering angle of the vehicle at time t;

[0030] Define the reward function R as:

[0031] Among them, d t Represents the distance between the vehicle's current position and the target parking space, defined as The vehicle's current location With parking space target P t =(P x ,P y ), collision t Indicates whether a collision occurs at time t. If a collision occurs, it is 1, otherwise it is 0, alignmenterror t Represents the error between the current direction of the vehicle and the target parking direction, defined as where θ Pis the orientation of the target parking space, w1, w2, w3 are the corresponding weights;

[0032] The parking path adjustment method in step S2 also includes:

[0033] Using a deep Q network to learn the optimal strategy, the agent's strategy π is maximized by maximizing the cumulative reward G t Get the optimal path planning strategy: Among them, Q(s t ,a t ; θ) is the state s t and action a t The Q value under the current network parameters θ, r t is the immediate reward at time step t, derived from the reward function R t , γ is the discount factor, θ is the weight parameter in the neural network,

[0034] The input of the neural network is the state s t , output is the Q value of each action, select the action with the largest Q value to guide the vehicle's next action;

[0035] The agent's strategy π is defined as t Next, the agent chooses action a t The probability distribution of strategy π selects actions by maximizing the Q value function, Q value function Q(s t ,a t ; θ) represents the state s t Select action a t The expected value of the future cumulative reward when the agent is in state s t When you select Action Maximizing Q:

[0036] Each action selected It is determined by maximizing the Q value in the current state; the cumulative reward G t is the immediate reward r that the agent receives from the environment at time t t and the sum of discounted future rewards for all subsequent time steps, G t =r t +γ·G t+1 , where r t is the immediate reward of the agent at time t, and γ is the discount factor, which takes a value between [0,1].

[0037] Furthermore, the parking path adjustment method in step S2 also includes:

[0038] Use the experience replay mechanism to store the experience in the memory pool, and the network randomly extracts experience for training each time:

[0039] in, is the loss function used to optimize the network weights θ, θ - is the weight parameter of the target network, used to stabilize training, E (s,a,r,s′) represents the expected action, r here is the immediate reward, is the discounted future reward, Q(s,a;θ) is the Q value of the current state; by minimizing the loss function, the reinforcement learning agent continuously adjusts the neural network parameter θ to learn the optimal strategy;

[0040] Real-time transmission of the latest status t To the neural network, dynamically adjust the path according to changes in the current environment:

[0041] in, represents the optimal action selected by the agent at time step t, that is, the next moving direction of the vehicle;

[0042] The vehicle chooses the best action Execute the corresponding move and obtain the new state s at the next time step t+1 t+1 , repeat the parking path adjustment steps until the vehicle successfully parks at the target parking space.

[0043] Furthermore, the relative position refers to the relative position between the same model of vehicle and the charging coil. The relative position is monitored by lidar and camera to ensure that the vehicle is completely aligned with the charging coil to avoid a decrease in charging efficiency due to misalignment of the parking position;

[0044] After the parking route in step S2 is confirmed, the relative position of the vehicle is continuously monitored during the charging process. If it is monitored that the vehicle is slightly displaced due to ground problems or other external factors, fine-tuning is performed simultaneously.

[0045] Furthermore, in step S2, the charging coils of the vehicle and the parking space are aligned as follows:

[0046] Collect historical charging data of vehicles of the same model in the parking lot, including:

[0047] The vehicle's historical parking location coordinates Vehicle heading angle The location of the charging coil (x c ,y c ) and the distance d between the parking space and the charging coil i , define the dataset Where N represents the total number of historical data, then for each vehicle i, at time step t, the relative position of the vehicle and the charging coil is: Among them, r i represents the position distance of vehicle i relative to the charging coil, represents the two-dimensional position coordinates of vehicle i when it is parked, (x c ,y c ) represents the fixed position of the charging coil;

[0048] The alignment error between the vehicle's heading angle and the charging coil position is: Among them, Δθ i Indicates the orientation error between the vehicle and the charging coil. represents the orientation angle of vehicle i when it is parked, θ c represents the fixed orientation angle of the charging coil, and Δθ i As close to 0 as possible, so that the vehicle and the charging coil are in the same direction;

[0049] Using the relative position and alignment error in the historical data, a Gaussian mixture model is used for modeling. Assuming that each parking mode k follows a Gaussian distribution, then:

[0050] Where P(r,Δθ|k) represents the joint probability of the relative position r and the orientation error Δθ between the vehicle and the charging coil in parking mode k, μ k represents the average stopping distance under parking mode k, σ k represents the distance standard deviation of parking pattern k, and the center position μ in the parking pattern k As the best parking position for the vehicle in this mode;

[0051] The method of aligning the charging coil of the vehicle and the parking space in step S2 also includes:

[0052] The optimal parking mode is expressed by minimizing the parking distance r and the orientation error Δθ, that is, Among them, k * represents the optimal parking mode, λ represents the error parameter that weighs the parking distance and orientation, and selects parking mode k * After that, the average stopping distance using this mode and orientation as the predicted value of the best parking space.

[0053] Furthermore, in step S2, during the actual parking process, when the vehicle approaches the parking space in real time, it is necessary to fine-tune the vehicle and dynamically adjust the vehicle position. and orientation The parking position is adjusted using a fine-tuning algorithm based on gradient descent, and the fine-tuning formula is:

[0054] Among them, α represents the learning rate, which is used to control the adjustment step size. represents the loss function, which is defined as the weighted sum of the relative position and orientation errors between the vehicle and the charging coil, that is, Here represents the distance error between the vehicle and the charging coil at time t, and β is a weight parameter used to adjust and orientation error The importance of the distance error of the charging coil is minimized by the loss function The vehicle gradually approaches the optimal parking position. represents partial derivative;

[0055] The parking position of the vehicle is analyzed and predicted based on historical data, and the parking position of the vehicle is adjusted in real time to align it precisely with the charging coil for an efficient wireless charging experience.

[0056] Furthermore, the charging power distribution method in step S3 is:

[0057] After the vehicle is parked, the health status data of the vehicle battery is obtained through the battery management system BMS;

[0058] Select the charging mode based on the obtained battery health status:

[0059] Set thresholds to distinguish between excellent, good, and poor battery health status, and high, medium, and low temperatures. When the battery health status is good and the temperature is moderate, select a high charging power.

[0060] When the battery temperature is high or the battery health status is poor, switch to low power charging mode.

[0061] Furthermore, the parking area is divided into an entry area, an exit area and a parking area. When parking, the user parks the car in the entry area or drives into the parking area by himself. When picking up the car, the user picks up the car from the exit area or the parking area. The parking area is used for parking and charging.

[0062] The beneficial effects of the present invention are:

[0063] The present invention effectively predicts the best parking position for vehicles of the same model through historical charging data analysis, and monitors the relative position of the vehicle in real time, accurately aligning the vehicle with the charging coil. Even if there is a slight shift, the system can automatically make fine adjustments to avoid manual repeated adjustments, thereby greatly improving the charging efficiency.

[0064] The present invention introduces deep reinforcement learning, so that vehicles can plan paths in real-time dynamic environments, flexibly adjust parking routes, and avoid interference from other vehicles or obstacles. The adaptive capability improves the efficiency of the system in complex scenarios and reduces the need for vehicles to repeatedly adjust paths.

[0065] The present invention monitors the battery health status in real time through the battery management system BMS, and dynamically adjusts the charging power according to the temperature and charging cycle number data. For batteries with poor health status or high temperature, the system will automatically select a lower power charging mode, thereby extending the battery life, preventing overheating, and improving the overall charging safety.

[0066] The present invention dynamically dispatches vehicles according to the traffic volume and charging demand. When the vehicle is fully charged, the system automatically moves the vehicle to the outbound area or the standby parking area, thereby improving the utilization efficiency of the charging piles and reducing the waiting time of other vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0068] Figure 1 This is a schematic flow chart of the automatic parking charging method of the present invention;

[0069] Figure 2 It is a schematic diagram of the structure of the automatic parking charging system of the present invention. DETAILED DESCRIPTION

[0070] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0071] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0072] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0073] Example 1, reference Figure 2 , this embodiment provides an automatic parking charging system, including:

[0074] A sensor network, which is used to collect environmental data in the parking lot in real time. The network consists of lidar, cameras, and ultrasonic radar;

[0075] A battery management BMS system data acquisition unit, which acquires the vehicle battery status through the vehicle's BMS system;

[0076] A path planning module that adjusts the parking path and aligns the vehicle with the charging coil in real time based on lidar and camera data;

[0077] The charging dynamic fine-tuning module is used to fine-tune the relative position of the vehicle and the charging coil during the charging phase. If the vehicle is slightly displaced, the vehicle position is adjusted to keep it aligned with the charging coil.

[0078] Charging power regulation module, which adjusts the charging power according to the battery status of the vehicle

[0079] The intelligent outbound module is used to move the charged vehicles to the outbound area.

[0080] Example 2, reference Figure 1 This embodiment provides an automatic parking charging method, comprising the following steps: Step S1, collecting data inside the parking lot,

[0081] A sensor network is deployed in the parking lot to collect environmental data in real time, and obtain the vehicle battery status through the vehicle battery management system BMS;

[0082] Environmental data includes vehicle location and movement, parking space occupancy, charging station status, and path obstruction;

[0083] Vehicle battery status includes battery temperature, remaining power, and number of charging cycles;

[0084] Specifically, by collecting data from inside the parking lot, the automatic parking system can reduce path planning errors caused by obstacles or other vehicle movements, thereby improving the system's response speed in complex environments.

[0085] Step S2: Parking path confirmation.

[0086] Based on environmental data, a dynamic path planning algorithm based on deep reinforcement learning is used to adjust the parking path. The environmental data is real-time environmental data, including path blockages caused by the movement of other vehicles;

[0087] At the same time, the historical charging relative position behavior pattern analysis of vehicles of the same model is carried out to predict the best parking spot for the vehicle and align the charging coil of the vehicle with the parking space;

[0088] The parking path adjustment method in step S2 is:

[0089] The parking environment is modeled as a state space S, where each state s t ∈S, at time step t, in, represents the current position coordinates of the vehicle at time t, Indicates the vehicle's heading angle. represents the obstacle coordinates, O t represents obstacles in the parking lot, P t represents the position coordinates of the target parking space, P t It is fixed during parking;

[0090] Define the vehicle's action space A, which is the set of actions performed by the vehicle at each time step, a t ∈A,a t =(Δv t ,Δθ t ), where Δv t represents the change in vehicle speed at time t, Δθ t represents the change in the steering angle of the vehicle at time t;

[0091] Define the reward function R as:

[0092] Among them, d t Represents the distance between the vehicle's current position and the target parking space, defined as The vehicle's current location With parking space target P t =(P x ,P y ), collision t Indicates whether a collision occurs at time t. If a collision occurs, it is 1, otherwise it is 0, alignmenterror t Represents the error between the current direction of the vehicle and the target parking direction, defined as where θ P is the orientation of the target parking space, w1, w2, w3 are the corresponding weights;

[0093] The parking path adjustment method in step S2 also includes:

[0094] Using a deep Q network to learn the optimal strategy, the agent's strategy π is maximized by maximizing the cumulative reward G t Get the optimal path planning strategy: Among them, Q(s t ,a t ; θ) is the state s t and action a t The Q value under the current network parameters θ, r t is the immediate reward at time step t, derived from the reward function R t , γ is the discount factor, θ is the weight parameter in the neural network,

[0095] The input of the neural network is the state s t , output is the Q value of each action, select the action with the largest Q value to guide the vehicle's next action;

[0096] The agent's strategy π is defined as t Next, the agent chooses action a t The probability distribution of strategy π selects actions by maximizing the Q value function, Q value function Q(s t ,a t ; θ) represents the state s t Select action a t The expected value of the future cumulative reward when the agent is in state s t When you select Action Maximizing Q:

[0097] Each action selected It is determined by maximizing the Q value in the current state; the cumulative reward G t is the immediate reward r that the agent receives from the environment at time t t and the sum of discounted future rewards for all subsequent time steps, G t =r t +γ·G t+1 , where r t is the immediate reward of the agent at time t, γ is the discount factor, and its value is between [0,1];

[0098] The parking path adjustment method in step S2 also includes:

[0099] Use the experience replay mechanism to store the experience in the memory pool, and the network randomly extracts experience for training each time:

[0100] in, is the loss function used to optimize the network weights θ, θ - is the weight parameter of the target network, used to stabilize training, E (s,a,r,s′) represents the expected action, r here is the immediate reward, is the discounted future reward, Q(s,a;θ) is the Q value of the current state; by minimizing the loss function, the reinforcement learning agent continuously adjusts the neural network parameter θ to learn the optimal strategy;

[0101] Real-time transmission of the latest status t To the neural network, dynamically adjust the path according to changes in the current environment:

[0102] in, represents the optimal action selected by the agent at time step t, that is, the next moving direction of the vehicle;

[0103] The vehicle chooses the best action Execute the corresponding move and obtain the new state s at the next time step t+1 t+1 , repeat the parking path adjustment steps until the vehicle successfully parks at the target parking space;

[0104] Specifically, path planning based on deep reinforcement learning (DRL) greatly improves the intelligence level of the system. Unlike traditional fixed path planning algorithms, deep reinforcement learning enables the system to autonomously learn the optimal path planning strategies in complex scenarios in continuous interaction. Through the deep Q network, the system can maximize the accumulated rewards and intelligently select the optimal actions, thereby achieving efficient and safe path planning. The deep Q network is also combined with the experience replay mechanism, allowing the system to learn and improve from past mistakes, so that the efficiency of each operation is gradually improved.

[0105] Relative position refers to the relative position between the same model of vehicle and the charging coil. The relative position is monitored by lidar and camera to ensure that the vehicle is completely aligned with the charging coil to avoid a decrease in charging efficiency due to misaligned parking positions.

[0106] After the parking route in step S2 is confirmed, the relative position of the vehicle is continuously monitored during the charging process. If the vehicle is slightly displaced due to ground problems or other external factors, fine-tuning is performed simultaneously;

[0107] In step S2, the charging coil alignment between the vehicle and the parking space is as follows:

[0108] Collect historical charging data of vehicles of the same model in the parking lot, including:

[0109] The vehicle's historical parking location coordinates Vehicle heading angle The location of the charging coil (x c ,y c ) and the distance d between the parking space and the charging coil i , define the dataset Where N represents the total number of historical data, then for each vehicle i, at time step t, the relative position of the vehicle and the charging coil is: Among them, r i represents the position distance of vehicle i relative to the charging coil, represents the two-dimensional position coordinates of vehicle i when it is parked, (x c ,y c ) represents the fixed position of the charging coil;

[0110] The alignment error between the vehicle's heading angle and the charging coil position is: Among them, Δθ i Indicates the orientation error between the vehicle and the charging coil. represents the orientation angle of vehicle i when it is parked, θ c represents the fixed orientation angle of the charging coil, and Δθ i As close to 0 as possible, so that the vehicle and the charging coil are in the same direction;

[0111] Using the relative position and alignment error in the historical data, a Gaussian mixture model is used for modeling. Assuming that each parking mode k follows a Gaussian distribution, then:

[0112] Where P(r,Δθ|k) represents the joint probability of the relative position r and the orientation error Δθ between the vehicle and the charging coil in parking mode k, μ k represents the average stopping distance under parking mode k, σ k represents the distance standard deviation of parking pattern k, and the center position μ in the parking pattern k As the best parking position for the vehicle in this mode;

[0113] The method of aligning the charging coil of the vehicle and the parking space in step S2 also includes:

[0114] The optimal parking mode is expressed by minimizing the parking distance r and the orientation error Δθ, that is, Among them, k * represents the optimal parking mode, λ represents the error parameter that weighs the parking distance and orientation, and selects parking mode k * After that, the average stopping distance using this mode and orientation As the predicted value of the best parking space;

[0115] In step S2, during the actual parking process, when the vehicle approaches the parking space in real time, it is necessary to fine-tune the vehicle and dynamically adjust the vehicle position. and orientation The parking position is adjusted using a fine-tuning algorithm based on gradient descent, and the fine-tuning formula is:

[0116] Among them, α represents the learning rate, which is used to control the adjustment step size. represents the loss function, which is defined as the weighted sum of the relative position and orientation errors between the vehicle and the charging coil, that is, Here represents the distance error between the vehicle and the charging coil at time t, and β is a weight parameter used to adjust and orientation error The importance of the distance error of the charging coil is minimized by the loss function The vehicle gradually approaches the optimal parking position. represents partial derivative;

[0117] Analyze and predict the parking position of vehicles based on historical data, and adjust the parking position of vehicles in real time to align them precisely with the charging coil for an efficient wireless charging experience.

[0118] Specifically, in terms of the alignment of the vehicle and the charging coil, by analyzing the historical data of vehicles of the same model, the Gaussian mixture model is used to model and predict the optimal parking position of the vehicle. Based on historical patterns, the most suitable parking position is predicted to ensure the precise alignment of the vehicle and the charging coil to the greatest extent. Compared with the traditional parking alignment method, the concept of orientation error is introduced to ensure that the vehicle and the charging coil are also completely aligned in direction, further improving the charging efficiency. In addition, the real-time fine-tuning algorithm ensures that even if the vehicle is slightly offset, the position can be dynamically adjusted to avoid a decrease in charging efficiency or parking failure.

[0119] Step S3, charging power distribution,

[0120] After the vehicle is parked according to the path determined in step S2, the battery health status data obtained from the battery management system BMS is analyzed, and the charging mode and power of the charging pile are adjusted according to the battery status; the battery is prevented from aging or overheating during charging, and a lower power is automatically allocated to the battery in poor health status to ensure charging safety;

[0121] The charging power distribution method in step S3 is:

[0122] After the vehicle is parked, the health status data of the vehicle battery is obtained through the battery management system BMS;

[0123] Select the charging mode based on the obtained battery health status:

[0124] Set thresholds to distinguish between excellent, good, and poor battery health status, and high, medium, and low temperatures. When the battery health status is good and the temperature is moderate, select a high charging power.

[0125] When the battery temperature is high or the battery health status is poor, switch to low-power charging mode;

[0126] Specifically, in terms of charging power distribution, through real-time communication with the vehicle's battery management system BMS, the battery health status is obtained, and the charging power is dynamically adjusted according to the battery's current temperature, remaining power, and number of charging cycles, effectively preventing battery damage caused by battery overheating or aging.

[0127] Step S4: Parking space management after charging is completed.

[0128] After charging is completed, the vehicle's stay time is managed according to the parking lot's demand and traffic flow. If there are other vehicles waiting to be charged, the vehicles that have completed charging will be arranged to move to the outbound area;

[0129] The parking area is divided into the entry area, exit area and parking area. When parking, users park their cars in the entry area or drive into the parking area. When picking up the car, users pick up the car from the exit area or parking area. The parking area is used for parking and charging.

[0130] Specifically, the parking position of vehicles is intelligently dispatched according to the charging status and traffic flow. Vehicles that have completed charging can vacate the charging pile position in time for use by other vehicles, reducing the user's waiting time and improving the user experience.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An automatic parking charging system, Features: include, A sensor network, which is used to collect environmental data in the parking lot in real time. The network consists of lidar, cameras, and ultrasonic radar; A battery management BMS system data acquisition unit, which acquires the vehicle battery status through the vehicle's BMS system; A path planning module that adjusts the parking path and aligns the vehicle with the charging coil in real time based on lidar and camera data; The charging dynamic fine-tuning module is used to fine-tune the relative position of the vehicle and the charging coil during the charging phase. If the vehicle is slightly displaced, the vehicle position is adjusted to keep it aligned with the charging coil. Charging power regulation module, which adjusts the charging power according to the battery status of the vehicle The intelligent outbound module is used to move the charged vehicles to the outbound area.

2. An automatic parking charging method, It is characterized in that An automatic parking charging system according to claim 1 comprises the following steps: Step S1, data collection inside the parking lot. A sensor network is deployed in the parking lot to collect environmental data in the parking lot in real time, and obtain the vehicle battery status through the vehicle battery management BMS system; Step S2, parking path confirmation, based on environmental data, a dynamic path planning algorithm based on deep reinforcement learning is used to adjust the parking path, and the environmental data is real-time environmental data, including path obstruction caused by the movement of other vehicles; At the same time, the historical charging relative position behavior pattern analysis of the same model vehicles is carried out to align the charging coils of the vehicles with the parking spaces. Step S3, charging power allocation, after the vehicle is parked according to the path determined in step S2, the battery health status data obtained from the battery management system BMS is analyzed, and the charging mode and power of the charging pile are adjusted according to the battery status; Step S4, after charging is completed, parking space management is performed. After charging is completed, the vehicle's stay time is managed according to the parking lot demand and traffic flow. If there are other vehicles waiting to be charged, the vehicles that have completed charging are arranged to be moved to the outbound area.

3. An automatic parking charging method according to claim 2, It is characterized in that Environmental data includes vehicle location and movement, parking space occupancy, charging station status, and path obstruction; The vehicle battery status includes battery temperature, remaining power, and number of charging cycles.

4. The automatic parking charging method according to claim 3, It is characterized in that The parking path adjustment method in step S2 is: The parking environment is modeled as a state space S, where each state s t ∈S, at time step t, in, represents the current position coordinates of the vehicle at time t, Indicates the vehicle's heading angle. represents the obstacle coordinates, O t represents obstacles in the parking lot, P t represents the position coordinates of the target parking space, P t It is fixed during parking; Define the vehicle's action space A, which is the set of actions performed by the vehicle at each time step, a t ∈A,a t =(Δv t ,Δθ t ), where Δv t represents the speed change of the vehicle at time t, Δθ t represents the change in the steering angle of the vehicle at time t; Define the reward function R as: Among them, d t Represents the distance between the vehicle's current position and the target parking space, defined as The vehicle's current location With parking space target P t =(P x ,P y ), collision t Indicates whether a collision occurs at time t. If a collision occurs, it is 1, otherwise it is 0, alignmenterror t Represents the error between the current direction of the vehicle and the target parking direction, defined as where θ P is the orientation of the target parking space, w 1 ,w 2 ,w 3 is the corresponding weight; The parking path adjustment method in step S2 also includes: Using a deep Q network to learn the optimal strategy, the agent's strategy π is maximized by maximizing the cumulative reward G t Get the optimal path planning strategy: Among them, Q(s t ,a t ; θ) is the state s t and action a t The Q value under the current network parameters θ, r t is the immediate reward at time step t, derived from the reward function R t , γ is the discount factor, θ is the weight parameter in the neural network, The input of the neural network is the state s t , output is the Q value of each action, select the action with the largest Q value to guide the vehicle's next action; The agent's strategy π is defined as t Next, the agent chooses action a t The probability distribution of strategy π selects actions by maximizing the Q value function, Q value function Q(s t ,a t ; θ) represents the state s t Select action a t The expected value of the future cumulative reward when the agent is in state s t When you select Action Maximizing Q: Each action selected It is determined by maximizing the Q value in the current state; the cumulative reward G t is the immediate reward r that the agent receives from the environment at time t t and the sum of discounted future rewards for all subsequent time steps, G t =r t +γ·G t+1 , where r t is the immediate reward of the agent at time t, and γ is the discount factor, which takes a value between [0,1].

5. An automatic parking charging method according to claim 4, It is characterized in that The parking path adjustment method in step S2 also includes: Use the experience replay mechanism to store the experience in the memory pool, and the network randomly extracts experience for training each time: in, is the loss function used to optimize the network weights θ, θ - is the weight parameter of the target network, used to stabilize the training, E (s,a,r,s′) represents the expected action, r here is the immediate reward, is the discounted future reward, Q(s,a;θ) is the Q value of the current state; Real-time transmission of the latest status t To the neural network, dynamically adjust the path according to changes in the current environment: in, represents the optimal action selected by the agent at time step t, that is, the next moving direction of the vehicle; The vehicle chooses the best action Execute the corresponding move and obtain the new state s at the next time step t+1 t+1 , repeat the parking path adjustment steps until the vehicle successfully parks at the target parking space.

6. An automatic parking charging method according to claim 5, It is characterized in that The relative position refers to the relative position between vehicles of the same model and the charging coil. The relative position is monitored by lidar and cameras. After the parking route in step S2 is confirmed, the relative position of the vehicle is continuously monitored during the charging process. If it is detected that the vehicle is slightly displaced due to ground problems or other external factors, fine-tuning is performed simultaneously.

7. An automatic parking charging method according to claim 6, It is characterized in that In step S2, the charging coil alignment between the vehicle and the parking space is as follows: Collect historical charging data of vehicles of the same model in the parking lot, including: The vehicle's historical parking location coordinates Vehicle heading angle The location of the charging coil (x c ,y c ) and the distance d between the parking space and the charging coil i , define the dataset Where N represents the total number of historical data, then for each vehicle i, at time step t, the relative position of the vehicle and the charging coil is: Among them, r i represents the position distance of vehicle i relative to the charging coil, represents the two-dimensional position coordinates of vehicle i when it is parked, (x c ,y c ) represents the fixed position of the charging coil; The alignment error between the vehicle's heading angle and the charging coil position is: Among them, Δθ i Indicates the orientation error between the vehicle and the charging coil. represents the orientation angle of vehicle i when it is parked, θ c represents the fixed orientation angle of the charging coil, and Δθ i As close to 0 as possible, so that the vehicle and the charging coil are in the same direction; Using the relative position and alignment error in the historical data, a Gaussian mixture model is used for modeling. Assuming that each parking mode k follows a Gaussian distribution, then: Where P(r,Δθ|k) represents the joint probability of the relative position r and the orientation error Δθ between the vehicle and the charging coil in parking mode k, μ k represents the average stopping distance under parking mode k, σ k represents the distance standard deviation of parking pattern k, and the center position μ in the parking pattern k As the best parking position for the vehicle in this mode; The method of aligning the charging coil of the vehicle and the parking space in step S2 also includes: The optimal parking mode is expressed by minimizing the parking distance r and the orientation error Δθ, that is, Among them, k * represents the optimal parking mode, λ represents the error parameter that weighs the parking distance and orientation, and selects parking mode k * After that, the average stopping distance using this mode and orientation as the predicted value of the best parking space.

8. An automatic parking charging method according to claim 7, It is characterized in that In step S2, during the actual parking process, when the vehicle approaches the parking space in real time, it is necessary to fine-tune the vehicle and dynamically adjust the vehicle position. and orientation The parking position is adjusted using a fine-tuning algorithm based on gradient descent, and the fine-tuning formula is: Among them, α represents the learning rate, which is used to control the adjustment step size. represents the loss function, which is defined as the weighted sum of the relative position and orientation errors between the vehicle and the charging coil, that is, Here represents the distance error between the vehicle and the charging coil at time t, and β is a weight parameter used to adjust and orientation error The importance of the distance error of the charging coil is minimized by the loss function The vehicle gradually approaches the optimal parking position. represents partial derivative.

9. An automatic parking charging method according to claim 8, It is characterized in that The charging power distribution method in step S3 is: After the vehicle is parked, the health status data of the vehicle battery is obtained through the battery management system BMS; Select the charging mode based on the obtained battery health status: Set thresholds to distinguish between excellent, good, and poor battery health status, and high, medium, and low temperatures. When the battery health status is good and the temperature is moderate, select a high charging power. When the battery temperature is high or the battery health status is poor, switch to low power charging mode.

10. An automatic parking charging method according to claim 9, It is characterized in that The parking area is divided into an entry area, an exit area and a parking area. When parking, the user parks the car in the entry area or drives into the parking area by himself. When picking up the car, the user picks up the car from the exit area or the parking area. The parking area is used for parking and charging.

Citation Information

Patent Citations

  • Wireless charging coil alignment system based on automatic parking technology

    CN109733216A

  • Parking method, device, vehicle, apparatus and medium

    CN110097779A

  • Parking path planning method based on reinforcement learning

    CN111098852A

  • Unmanned vehicle adaptive path planning method based on dynamic window method and near-end strategy

    CN116679719A

  • Parking space intelligent dynamic distribution and guidance method, device, equipment and medium

    CN116704808A

Cited By

  • Charging parking space dynamic management solution method and system based on intelligent hardware integration

    CN121032124A