Battery car charging integrated management system based on edge calculation

Abstract

The invention relates to the technical field of battery car charging management, in particular to a battery car charging integrated management system based on edge calculation, which acquires battery parameters of a battery car and state parameters of a charging pile in real time through a battery car end acquisition module and a charging pile end acquisition module, and performs local processing through an edge calculation unit. The real-time performance of data processing is improved, and abnormal conditions can be quickly responded. Charging behaviors are analyzed through deep learning, abnormal conditions are judged, and data are distributed to a load balancing unit or a risk assessment unit. The load balancing unit utilizes a reinforcement learning model to optimize charging load distribution according to the power grid load, the battery car demand and the charging pile state so as to avoid power grid overload. The risk assessment unit establishes a model through historical data, monitors the risk in real time and gives an alarm, and meanwhile, through edge calculation and local deployment of a classification model, the dependence on the cloud is reduced, and the calculation efficiency and the management efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery vehicle charging management, and in particular to an integrated battery vehicle charging management system based on edge computing. Background Art

[0002] Battery cars have become one of the important tools for short-distance travel in cities. As an efficient vehicle parking solution, stereo garages have gradually been widely used. Stereo garages store vehicles through multi-layer mechanical structures, effectively saving ground space and greatly alleviating parking pressure in cities. However, with the increase in the number of battery car users, how to reasonably, safely and efficiently charge battery cars in stereo garages has become an urgent problem to be solved.

[0003] Traditional battery car charging methods are mostly concentrated in flat parking lots or independent charging stations. The layout of charging piles in these scenarios is relatively simple, and the charging load management and safety issues are also relatively simple. However, in the stereo garage scenario, battery car charging involves charging safety management in multiple layers of space. Due to the dense concentration of vehicles in the stereo garage, the power load of battery car charging piles fluctuates greatly, and the impact on the grid load is more complicated when concentrated charging of battery cars occurs. If the power load cannot be reasonably distributed or abnormal conditions are not discovered in time, it is very easy to cause safety hazards such as power overload, equipment damage, and even fire. Summary of the invention

[0004] To solve the above problems, the present invention provides an electric vehicle battery charging integrated management system based on edge computing. By combining edge computing with machine learning models, it realizes the intelligent management of large-scale electric vehicle battery charging and reduces the server computing power cost.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A battery vehicle charging integrated management system based on edge computing, comprising a battery vehicle end collection module, a charging pile end collection module and an intelligent management module, wherein the battery vehicle end collection module and the charging pile end collection module are respectively connected to the intelligent management module;

[0007] The battery vehicle end acquisition module is used to collect battery parameters in the battery vehicle and send them to the edge computing unit, including battery model, capacity, temperature, battery voltage and charging rate;

[0008] The charging pile end acquisition module is used to collect the state parameters of the charging pile load and each port of the charging pile that the power grid can support, and the state parameters of the port include output voltage, output current and load power, and send them to the edge computing unit;

[0009] The intelligent management module includes an edge computing unit, a load balancing unit and a risk assessment unit, and the edge computing unit is communicatively connected with the load balancing unit and the risk assessment unit respectively;

[0010] The edge computing unit is used to determine whether the current charging behavior is abnormal and determine the abnormality type according to the charging pile load, battery parameters and status parameters of each port supported by the power grid through deep learning, and distribute and send the charging pile load, battery parameters and status parameters of each port supported by the power grid to the load balancing unit or the risk assessment unit according to the abnormality type;

[0011] The load balancing unit is used to adjust the load according to the reinforcement learning model based on the charging pile load that the power grid can support, the battery parameters and the status parameters of each port;

[0012] The risk assessment unit is used to build a risk assessment model based on historical abnormality types, battery parameters and status parameters of each port, and output the risk level and issue an alarm prompt based on the real-time abnormality type, battery parameters and status parameters of each port through the risk assessment model.

[0013] Further, the edge computing unit includes a battery central control subunit and a charging pile central control subunit, and the battery central control subunit and the charging pile central control subunit are communicatively connected;

[0014] The battery central control subunit is used to receive battery parameters for standardization and send them to the charging pile central control subunit;

[0015] The charging pile central control subunit is used to perform the following steps:

[0016] The load of the charging pile supported by the power grid, the battery parameters and the state parameters of the port are used as the input of the feedforward neural network, and the abnormality type is output, which includes load abnormality and charging abnormality;

[0017] The battery parameters and the status parameters of each port are sent to the load balancing unit or the risk assessment unit according to the abnormality type.

[0018] Furthermore, the training step of the feedforward neural network includes:

[0019] Normalize the charging pile load, battery parameters and port status parameters that the power grid can support, and mark the abnormal types;

[0020] Each parameter of the normalized power grid-supported charging pile load, battery parameters, and port status parameters is split to construct input neurons;

[0021] The features of the input neurons are calculated and extracted layer by layer through the hidden layers in the feedforward neural network, and the activation function is used for nonlinear feature mapping, and the high-order features of the abnormal pattern are extracted and output through the output layer integration.

[0022] The loss function is calculated based on the output results of the output layer and the marked anomaly types, and the network weights are adjusted using the back propagation algorithm.

[0023] Further, the sending of the battery parameters and the status parameters of each port to the load balancing unit or the risk assessment unit according to the abnormality type includes:

[0024] If the abnormality type is load abnormality, the charging pile load that the power grid can support, the battery parameters and the status parameters of each port are sent to the load balancing unit;

[0025] If the abnormality type is charging abnormality, the battery parameters and the status parameters of each port are sent to the risk assessment unit.

[0026] Furthermore, the load balancing unit is used to perform the following steps:

[0027] The ratio of the charging pile load supported by the preset power grid to the available load of the charging pile is generated according to the real-time available load of the charging pile supported by the power grid;

[0028] The current total load of the charging pile is calculated according to the state parameters of the port. The current total load of the charging pile, the real-time available load of the charging pile, and the battery parameters of each port are taken as the state, and the load adjustment of the port is taken as the action to build a reinforcement learning model, and the load adjustment is output through the reinforcement learning model.

[0029] Furthermore, the formula of the reinforcement learning model is as follows:

[0030]

[0031] Where s is the state, including the battery parameters and the state parameters of the charging pile port; a is the action, including the load regulation parameters of all ports; Q(s, a) is the expected return that can be obtained after taking action a in state s; r is the immediate reward value after performing the regulation action; γ is the discount factor; is the execution action; s′ is the next state after performing action a; a′ is the optimal regulation action; max a′ Q(s′, a′) is the maximum expected reward that can be obtained by selecting the optimal adjustment action a′ in the next state s′.

[0032] Furthermore, the formula of the instant reward value r is as follows:

[0033] r=-α·|L current -L optimal |-β·Eloss ;

[0034] Among them, L current is the current load of the charging pile; L optimal E is the real-time available load of the charging pile; loss is the energy loss caused by load imbalance or power loss; α and β are the weight coefficients of load balance and energy loss respectively.

[0035] Furthermore, the risk assessment model is constructed according to the abnormality type, battery parameters and status parameters of each port in the history, including the following steps:

[0036] Based on the abnormal data in history, collect historical data sets including abnormal type, battery parameters and status parameters of each port, and mark the corresponding risk level for each record;

[0037] Divide the historical data set with completed risk level annotation into a training set and a test set;

[0038] Based on the decision tree algorithm, the parameters in the training set are recursively split into two elements, and the risk level is output through the leaf nodes to obtain the risk assessment model;

[0039] The risk assessment model is validated through the test set.

[0040] Furthermore, the risk assessment unit and the load balancing unit are both deployed on a cloud server.

[0041] The beneficial effect of the present invention is that the present invention collects the battery parameters of the battery vehicle and the state parameters of the charging pile in real time through the battery vehicle end collection module and the charging pile end collection module. These data are first processed locally by the edge computing unit, thereby reducing the pressure of cloud computing power and being able to quickly respond to abnormal situations in the charging process. The charging behavior of the battery vehicle is intelligently analyzed by a deep learning algorithm to determine whether an abnormality occurs in the charging process, and the data is distributed to the load balancing unit or the risk assessment unit according to the abnormality type. The load balancing unit is used to optimize the charging load distribution according to the available load of the power grid, the charging demand of the battery vehicle and the state of the charging pile, so as to avoid overload of the power grid during peak hours. At the same time, the risk assessment unit establishes a risk assessment model by training historical abnormal data, which can monitor the potential risks in the charging state in real time, and issue an alarm prompt according to the risk level to prevent accidents in advance. At the same time, the system uses edge computing to deploy the classification model locally for data classification processing, reducing the dependence on the computing power of the cloud server and effectively improving the computing management efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a structural schematic diagram of an electric battery vehicle charging integrated management system based on edge computing in the present invention.

[0043] Figure 2 It is a flow chart of the training steps of the feedforward neural network in the present invention. DETAILED DESCRIPTION

[0044] See also Figure 1-2 As shown, the present invention relates to an integrated battery vehicle charging management system based on edge computing, including a battery vehicle end collection module, a charging pile end collection module and an intelligent management module, wherein the battery vehicle end collection module and the charging pile end collection module are respectively connected to the intelligent management module;

[0045] The battery vehicle end acquisition module is used to collect battery parameters in the battery vehicle and send them to the edge computing unit, including battery model, capacity, temperature, battery voltage and charging rate;

[0046] The charging pile end acquisition module is used to collect the state parameters of the charging pile load and each port of the charging pile that the power grid can support, and the state parameters of the port include output voltage, output current and load power, and send them to the edge computing unit;

[0047] The intelligent management module includes an edge computing unit, a load balancing unit and a risk assessment unit, and the edge computing unit is communicatively connected with the load balancing unit and the risk assessment unit respectively;

[0048] The edge computing unit is used to determine whether the current charging behavior is abnormal and determine the abnormality type according to the charging pile load, battery parameters and status parameters of each port supported by the power grid through deep learning, and distribute and send the charging pile load, battery parameters and status parameters of each port supported by the power grid to the load balancing unit or the risk assessment unit according to the abnormality type;

[0049] The load balancing unit is used to adjust the load according to the reinforcement learning model based on the charging pile load that the power grid can support, the battery parameters and the status parameters of each port;

[0050] The risk assessment unit is used to build a risk assessment model based on historical abnormality types, battery parameters and status parameters of each port, and output the risk level and issue an alarm prompt based on the real-time abnormality type, battery parameters and status parameters of each port through the risk assessment model.

[0051] It should be noted that the battery vehicle end acquisition module, as the front-end data acquisition device of the system, mainly monitors the key parameters of the battery in real time through a series of sensors. The collected parameters include the battery model, capacity, temperature, battery voltage and charging rate. Each parameter has a dedicated sensor responsible for acquisition, which can include: voltage sensor (such as Hall effect sensor) is used to monitor the voltage change of the battery to ensure that the voltage fluctuation is within a safe range; temperature sensor (such as NTC thermistor) is responsible for collecting battery temperature to avoid overheating caused by excessive temperature; current sensor is used to detect changes in charging rate and evaluate charging efficiency and safety. All collected data are preliminarily processed and filtered by an embedded acquisition controller (such as STM32 microcontroller) to remove noise data, and the data is transmitted to the edge computing unit through a wireless communication module (such as ZigBee, LoRa, BLE and other low-power communication protocols). The charging pile end acquisition module is used to collect the status information of the charging pile and the power grid in real time to ensure that the charging process is carried out within the load range that the power grid can bear. Each charging pile port is equipped with a current and voltage sensor to collect the output voltage, output current and load power of each port in real time. The voltage sensor can accurately detect the output voltage to ensure its stability; the current sensor monitors the transient changes of the output current to detect whether there is a short circuit or overload; the power sensor is used to calculate the total power consumption of the charging pile to evaluate the load capacity of the power grid. The data of these sensors are processed in real time by the embedded processing unit (such as FPGA or DSP) built into the charging pile, and the results are passed to the edge computing unit of the system as the basis for adjusting the charging strategy. The intelligent management module is the core control unit of the entire system, which contains the edge computing unit, load balancing unit and risk assessment unit. The design of this module is based on an embedded computing platform, such as the NVIDIA Jetson series or Intel Movidius module, which can process a large amount of data locally in real time, reduce dependence on remote cloud computing, and improve response speed and data privacy. In the intelligent management module, the edge computing unit is responsible for receiving and integrating data from the battery vehicle and the charging pile. Through deep learning algorithms (such as convolutional neural networks), the unit analyzes the battery status data and the load data of the charging pile in real time to detect abnormal behavior during charging. The deep learning model is pre-trained on a cloud server database containing a large amount of battery and charging data. It can accurately identify different types of anomalies (such as abnormal battery temperature, excessive charging rate, voltage fluctuations, etc.), and then deploy the trained model locally on the edge computing device. When abnormal behavior is detected, the edge computing unit labels and classifies the abnormal information, and sends the data to the load balancing unit or risk assessment unit for further processing based on the specific type of anomaly. The load balancing unit realizes the intelligent allocation of charging power based on the reinforcement learning model to optimize the load balance between the power grid and the charging pile.Reinforcement learning models (such as deep Q learning algorithms DQN or policy gradient algorithms) learn how to best allocate charging power under different charging conditions by training charging history data. The core task of this unit is to dynamically adjust the charging output of each port based on real-time battery parameters and the status of the charging pile. For example, when it is detected that the load power of a port is too high, the system will automatically reduce the output power of the port and redistribute the excess power to the port with lower load, thereby avoiding local overload of the charging pile. The risk assessment unit is deployed through the cloud server, triggers the service support of the edge computing unit through cloud instructions, and uses machine learning algorithms to predict and evaluate the potential risks in the charging process in real time. The unit builds a risk assessment model based on the anomaly type, battery parameters and charging pile status parameters in the historical data. The risk assessment model is based on random forest or gradient boosted decision tree (GBDT) algorithms, which can extract abnormal patterns from a large amount of historical data and classify and predict them. During the real-time monitoring process, when the system detects a new abnormal signal, the risk assessment unit outputs the corresponding risk level by inputting the current data into the model and triggers the alarm mechanism. For example, when the system determines that an abnormal behavior may cause battery overheating or charging pile failure, it will immediately issue a high-risk warning and display detailed risk information on the charging pile LED screen. At the same time, it will notify relevant maintenance personnel through a buzzer or remote communication module (such as 4G / 5G network) to take emergency measures.

[0052] Further, the edge computing unit includes a battery central control subunit and a charging pile central control subunit, and the battery central control subunit and the charging pile central control subunit are communicatively connected;

[0053] The battery central control subunit is used to receive battery parameters for standardization and send them to the charging pile central control subunit;

[0054] The charging pile central control subunit is used to perform the following steps:

[0055] The load of the charging pile supported by the power grid, the battery parameters and the state parameters of the port are used as the input of the feedforward neural network, and the abnormality type is output, which includes load abnormality and charging abnormality;

[0056] The battery parameters and the status parameters of each port are sent to the load balancing unit or the risk assessment unit according to the abnormality type.

[0057] In some embodiments, the battery central control subunit is mainly responsible for receiving and preprocessing battery parameters. Battery parameters include battery model, capacity, temperature, battery voltage, and charging rate. These parameters are collected in real time by sensors embedded in the battery vehicle, and the data is transmitted to the battery central control subunit through a low-latency communication interface (such as a CAN bus or RS485 interface). The battery central control subunit standardizes these collected battery parameters through a built-in microcontroller (such as an ARM Cortex-M series). The purpose of standardization is to convert various parameters into a unified scale so that subsequent algorithms can accurately analyze them. For example, the battery temperature may be collected in degrees Celsius, and the battery voltage is in volts. Through standardization, these data are mapped to the same numerical range, reducing the impact between different dimensions and improving data consistency. The standardized battery parameters are sent to the charging pile central control subunit through a real-time communication protocol (such as MQTT or HTTP RESTful interface) for further anomaly detection and analysis. After receiving the standardized battery parameters from the battery central control subunit, the charging pile central control subunit combines the charging load information supported by the power grid and the status parameters of the charging pile port to execute the intelligent algorithm for abnormality detection. This process uses a feedforward neural network (FNN) for real-time analysis. The feedforward neural network is a basic deep learning model that is suitable for processing nonlinear relationships between input and output. After completing the detection and classification of abnormal types, the charging pile central control subunit sends the relevant parameter data to a specific subsequent module for processing according to the different abnormal types. If the system detects "load abnormality", the central control subunit sends the battery parameters and the status parameters of the port to the load balancing unit. The load balancing unit adjusts the power output of the charging pile through a reinforcement learning algorithm to avoid grid overload and ensure charging safety. If "charging abnormality" is detected, the relevant data will be transmitted to the risk assessment unit. The risk assessment unit uses the risk assessment model constructed based on the abnormal parameters to evaluate the risk level in the battery charging process in real time and issue an alarm prompt when necessary.

[0058] Furthermore, the training step of the feedforward neural network includes:

[0059] Normalize the charging pile load, battery parameters and port status parameters that the power grid can support, and mark the abnormal types;

[0060] Each parameter of the normalized power grid-supported charging pile load, battery parameters, and port status parameters is split to construct input neurons;

[0061] The features of the input neurons are calculated and extracted layer by layer through the hidden layers in the feedforward neural network, and the activation function is used for nonlinear feature mapping, and the high-order features of the abnormal pattern are extracted and output through the output layer integration.

[0062] The loss function is calculated based on the output results of the output layer and the marked anomaly types, and the network weights are adjusted using the back propagation algorithm.

[0063] In some embodiments, data preprocessing is first performed. In this step, the power grid normalizes the load of the charging pile, various parameters of the battery (such as temperature, voltage, capacity), and real-time status parameters of the charging pile port (such as output current, voltage, etc.). The goal of normalization is to scale the data to a uniform range (usually between 0 and 1), thereby eliminating the dimensional differences between different data and ensuring that the subsequent algorithm is not affected by the size of the eigenvalue when processing the data. For example, the battery voltage and battery temperature may vary greatly in the data range, but after normalization, they are all converted to the same scale to facilitate unified analysis of the model. At this time, the normalized data will also be marked with abnormal types as a reference target for training the model. After the data preparation is completed, the step of constructing input neurons begins. The input layer of the neural network consists of multiple input neurons, and each normalized data feature corresponds to a neuron. The input layer is responsible for receiving all standardized parameter data, such as battery temperature, battery voltage, charging load supported by the power grid, and status information of the charging pile port. Through this structural design, the input layer can provide comprehensive operating status data, covering the key factors affecting the charging process. After the input data enters the hidden layer, the feature extraction and nonlinear mapping process begins. The hidden layer consists of multiple neurons, each of which is connected to the neurons in the previous layer through weights. The neurons in each hidden layer perform weighted summation of the input features and are processed by activation functions to capture the complex nonlinear relationships in the data. For example, when the battery temperature is detected to rise suddenly, the hidden layer neurons will respond to this abnormal change and reflect it in the feature extraction. The choice of activation function, such as ReLU (rectified linear unit), can enhance the model's ability to recognize complex features and ensure the system's sensitivity to changes in the battery and charging pile status. Under the step-by-step mapping of multiple hidden layers, the model gradually extracts high-order abnormal pattern features, which can reflect abnormal behaviors that may be hidden in the input data. After the feature extraction is completed, the data enters the output layer for final integration. The output layer determines whether the data is abnormal based on the high-order features extracted by the hidden layer and classifies it as "load abnormality" or "charging abnormality". This classification is based on the probability results of the output layer neurons, and the model ultimately outputs the probability of each abnormal type. For example, if the system detects that the grid load increases abnormally and the battery temperature increases rapidly during a charging process, the output layer will be more inclined to mark this situation as "charging abnormality". In this way, the system can effectively classify different types of abnormalities. In order to improve the prediction accuracy of the network, the model also performs loss calculations and weight adjustments. The loss function is used to measure the difference between the predicted result and the actual labeled value. In this process, the cross entropy loss function is often used, which can quantify the classification error of the model. For example, if the network mistakenly classifies "load abnormality" as "charging abnormality", the loss function will give a specific loss value based on the degree of error.The system uses the back-propagation algorithm to adjust the weights and biases of the neural network according to the loss value. The back-propagation process involves calculating the gradient of each weight and updating the weights based on the gradient information so that the model is closer to the correct result in the next iteration.

[0064] Further, the sending of the battery parameters and the status parameters of each port to the load balancing unit or the risk assessment unit according to the abnormality type includes:

[0065] If the abnormality type is load abnormality, the charging pile load that the power grid can support, the battery parameters and the status parameters of each port are sent to the load balancing unit;

[0066] If the abnormality type is charging abnormality, the battery parameters and the status parameters of each port are sent to the risk assessment unit.

[0067] In some embodiments, load anomalies usually involve load problems of the power grid or charging piles, such as power grid load overload, excessive current or abnormal voltage fluctuations; charging anomalies are related to the state of the battery or the supply current and voltage, such as excessive battery temperature, abnormal voltage or too fast charging rate. After determining the type of anomaly, the system takes corresponding countermeasures according to different anomaly types and passes the data to the appropriate module for further processing. When the anomaly type is load anomaly, the system will focus on the parameters related to the power grid and charging piles. In this case, the system needs to conduct a detailed analysis of the charging pile load, battery parameters and status parameters of each charging port supported by the power grid to adjust the load balance of the charging pile. To achieve this goal, the system sends these parameter data to the load balancing unit. Specifically, the load capacity of the power grid, the current voltage and temperature of the battery, and the real-time output current and voltage of each port of the charging pile are all passed as input data to the load balancing unit. This unit will dynamically adjust the power distribution of the charging pile using a reinforcement learning algorithm based on the real-time data received. The reinforcement learning algorithm finds the optimal charging strategy based on the historical load conditions and current status of the power grid and the port through continuous iteration, thereby avoiding overload of the power grid. For example, when the system detects that the current of a charging port is too high, the load balancing unit will reduce the charging power of the port and allocate more power to other ports with lower loads to ensure the balance of the overall load. When the abnormality type is charging abnormality, the system will focus on the analysis of the battery status. This type of abnormality usually means that the battery itself may have potential risks, such as excessive temperature, excessive charging current, or abnormal fluctuations in battery voltage. To this end, the system will send the detailed parameters of the battery and the status parameters of each port to the risk assessment unit. The risk assessment unit uses a risk prediction model based on machine learning, which can identify potential risk factors in battery parameters by analyzing and learning historical data. For example, if the system detects that the temperature of the battery rises rapidly in a short period of time, or the battery voltage fluctuates unstably, the risk assessment unit will combine similar data in history to determine whether the current charging behavior may cause safety problems. The unit will conduct a comprehensive evaluation of the received battery parameters and port status, generate a risk level, and output specific risk prompts. For example, if the risk assessment model determines that the current charging is high risk, the system will immediately trigger an alarm to notify the user or maintenance personnel to take corresponding measures, such as suspending charging or reducing the charging rate, to ensure safety.

[0068] Furthermore, the load balancing unit is used to perform the following steps:

[0069] The ratio of the charging pile load supported by the preset power grid to the available load of the charging pile is generated according to the real-time available load of the charging pile supported by the power grid;

[0070] The current total load of the charging pile is calculated according to the state parameters of the port. The current total load of the charging pile, the real-time available load of the charging pile, and the battery parameters of each port are taken as the state, and the load adjustment of the port is taken as the action to build a reinforcement learning model, and the load adjustment is output through the reinforcement learning model.

[0071] In some embodiments, first, the system needs to preset the ratio of the power grid's supportable charging pile load to the power grid's disposable load. The power grid's supportable charging pile load represents the maximum load capacity that the power grid can provide to the charging pile in the current state, while the power grid's disposable load represents the load range that the charging pile can flexibly adjust under this load capacity. In order to ensure that the charging pile is not overloaded, the system usually sets a safety factor based on the stability of the power grid to represent the proportional relationship between the power grid's disposable load and the power grid's supportable charging pile load. For example, if the safety factor is set to 0.8, it means that the power grid's disposable load is 80% of the power grid's maximum supported load. In actual operation, when the power grid state changes, the system generates the real-time disposable load of the charging pile based on the safety factor. After obtaining the real-time charging pile's disposable load, the system calculates the current total load of the charging pile based on the state parameters of each port. The state parameters include the output voltage, output current, battery temperature, and battery voltage of each port. These parameters are monitored in real time by high-precision sensors on the port and sent to the load balancing unit for calculation. Through these data, the system can accurately determine the current load of each port, thereby deriving the total load of the charging pile. This total load represents the power currently consumed by all ports together, and the system will compare it with the real-time available load of the charging pile as the basis for the next decision. The decision of load balancing depends on the construction of the reinforcement learning model. The reinforcement learning model represents the state of the charging pile as a state space, including the current total load of the charging pile, the real-time available load, and the battery parameters of each port. Each state in the state space represents the operation of the charging pile at a certain moment. Based on this state, the goal of the system is to find the best load adjustment strategy to maintain the stability of the power grid and the charging pile. In the reinforcement learning model, the load adjustment of each port (such as adjusting the charging current or voltage) is regarded as the action of the model. The action space includes increasing or decreasing the charging power of the port, through which the system can adjust the load distribution of each port. The training process of the reinforcement learning model aims to minimize the instability of the load and maximize the efficiency of charging. To this end, the model uses a reward function to measure the effect of each action. The design of the reward function is the key to the entire model. It will evaluate the quality of each action based on the state change of the port, the stability of the power grid, and the charging efficiency. For example, if the adjusted port load stabilizes the total load of the charging pile within the available range and optimizes the battery charging efficiency, the model will give a high reward; on the contrary, if the adjusted load exceeds the available range or causes grid fluctuations, a penalty will be given. The specific expression of the reward function may involve a weighted combination of multiple factors, such as charging time, battery health status, and grid stability. Through repeated trials and experience accumulation, the model continuously adjusts the port load strategy to find the optimal load regulation solution.In this process, the reinforcement learning algorithm uses algorithms such as Q-learning or deep Q network (DQN) to optimize the model using experience replay and Q value update mechanism. Each load adjustment action will update the Q value to represent the long-term benefit of the state-action pair. As the training progresses, the model gradually converges and can output the optimal load adjustment plan under different power grid and charging pile conditions.

[0072] Furthermore, the formula of the reinforcement learning model is as follows:

[0073]

[0074] Where s is the state, including the battery parameters and the state parameters of the charging pile port; a is the action, including the load regulation parameters of all ports; Q(s, a) is the expected return that can be obtained after taking action a in state s; r is the immediate reward value after performing the regulation action; γ is the discount factor; is the execution action; s′ is the next state after performing action a; a′ is the optimal regulation action; max a′ Q(s′, a′) is the maximum expected reward that can be obtained by selecting the optimal adjustment action a′ in the next state s′.

[0075] Specifically, the state s represents all environmental information of the charging system at a specific point in time, including battery parameters and state parameters of the charging pile port. Battery parameters may include battery temperature, voltage, and charging state, while the state parameters of the charging pile port cover the output voltage, output current, and load conditions of each port. The combination of all these parameters constitutes the current state of the system, which is used to describe the charging state of the system at the current moment. The expected return Q(s, a) represents the cumulative benefit that the system can obtain in the long term after selecting action a in a specific state s. This value reflects the impact of the current action on charging efficiency and grid stability in the future. The expected return is the core goal of the reinforcement learning model. The system continuously adjusts the Q value through multiple trials and training to find the best adjustment strategy under different states. The immediate reward value r is the feedback obtained immediately after the system performs an action in a specific state. The setting of the reward value is usually based on the effect of the current charging behavior. For example, when the adjusted charging power reduces the battery temperature or improves the charging efficiency, the system will receive a positive reward; conversely, if the adjustment causes grid fluctuations or abnormal battery heating, the system will be punished. The design of the reward value needs to comprehensively consider charging safety and efficiency to guide the system to learn the best adjustment strategy. The discount factor γ is used to measure the value of future rewards. The discount factor is between 0 and 1 and is used to balance the weight of immediate rewards and long-term returns. A larger value of γ means that the system pays more attention to the long-term optimization effect, while a smaller value of γ emphasizes the current immediate benefits. For the charging management system, setting γ appropriately can ensure that the system optimizes the current charging behavior while also considering possible future impacts. The next state s′ is the new state that the system enters after executing action a. This new state depends on the impact of the action on the current charging system. For example, after adjusting the output current of a port, the temperature of the battery may change, and the load distribution of the charging pile will also be readjusted. The system will record this change and use it as the basis for the next round of decision-making. The optimal adjustment action a′ is the best adjustment strategy that the system can take in the next state s′. By searching and evaluating all possible actions, the system will select the optimal action that can maximize the expected return. The process of selecting the optimal action is based on the existing Q value calculation, and the system will continuously update and optimize these values ​​to ensure the rationality of each decision. The maximum expected return is the highest return that the system can obtain after selecting the optimal action a' in state s'. This parameter is used to update the Q value, helping the system to evaluate the long-term effect of each action. By continuously calculating the maximum expected return, the system can gradually identify the optimal load regulation strategy.

[0076] Furthermore, the formula of the instant reward value r is as follows:

[0077] r=-α·|L current -L optimal |-β·E loss ;

[0078] Among them, L current is the current load of the charging pile; L optimal E is the real-time available load of the charging pile; loss is the energy loss caused by load imbalance or power loss; α and β are the weight coefficients of load balance and energy loss respectively.

[0079] Specifically, L optimal The optimal load value is calculated based on the available load of the power grid and the charging demand of the battery vehicle, ensuring that the impact of excessive load on the power grid is avoided while meeting the charging efficiency of the battery vehicle. Parameter E loss Indicates the energy loss caused by load imbalance or power loss during charging. The two weight coefficients α and β correspond to the importance of load balance and energy loss respectively. The specific value depends on the system optimization goal and grid characteristics. In the algorithm implementation, the system first collects the battery parameters of the battery vehicle and the status parameters of the charging pile port through sensors, and calculates the current load L in real time. current The system simultaneously analyzes the available load data of the power grid and dynamically generates the optimal load value L based on the charging demand of the battery vehicle. optimal , at each time step, the system calculates the instant reward value r based on the above formula to evaluate the pros and cons of the current charging regulation strategy. current With L optinal If the difference between E and loss If it is higher, the impact of energy loss in the instant reward value will also increase. By increasing the penalty weight β of energy loss, the system will be prompted to control power distribution more effectively during the charging process and reduce energy waste. During each charging load adjustment process, the system uses the current instant reward value as feedback information for the reinforcement learning model to update the strategy. Specifically, the reinforcement learning model continuously adjusts the output power and optimizes the load regulation scheme by maximizing the cumulative reward. During the training phase, the system uses historical data to build an experience pool, and trains the model based on the instant reward value through the Q-learning or DQN (deep Q network) algorithm. The model will gradually learn how to choose the optimal load regulation action under different charging environments and grid load conditions, so as to make better decisions in future real-time charging scenarios.

[0080] Furthermore, the risk assessment model is constructed according to the abnormality type, battery parameters and status parameters of each port in the history, including the following steps:

[0081] Based on the abnormal data in history, collect historical data sets including abnormal type, battery parameters and status parameters of each port, and mark the corresponding risk level for each record;

[0082] Divide the historical data set with completed risk level annotation into a training set and a test set;

[0083] Based on the decision tree algorithm, the parameters in the training set are recursively split into two elements, and the risk level is output through the leaf nodes to obtain the risk assessment model;

[0084] The risk assessment model is validated through the test set.

[0085] Specifically, first, the system collects historical data to form a data set containing abnormality types, battery parameters (such as battery model, capacity, temperature, battery voltage, charging rate, etc.) and status parameters of each port (such as output voltage, output current, load power). In order to ensure the accuracy and effectiveness of model training, each historical record will be labeled with a corresponding risk level, and the risk level is set based on the specific abnormality type and its potential impact on the safety of the charging system. Through this annotation, the historical data set becomes a complete data set that can be used for supervised learning. In the data processing stage, the system divides the labeled historical data set into a training set and a test set. The training set is used to build the model, while the test set is used for subsequent verification and evaluation. Such a division can ensure that the model still has good predictive and generalization capabilities when facing unseen data. Then, the system uses the decision tree algorithm to process the data in the training set. When building the model, the decision tree algorithm recursively performs binary segmentation on each feature. Specifically, the algorithm selects a feature at each node and divides the data set into two parts based on a certain threshold of the feature. This segmentation process selects the optimal segmentation point based on indicators such as information gain, entropy or Gini coefficient. In this way, through recursive segmentation, a tree structure is finally generated, in which each leaf node corresponds to a specific risk level output. These leaf nodes reflect the combination relationship between battery parameters and port status parameters under different abnormal conditions, providing a basis for risk assessment in the charging process. After the model is built, in order to verify its accuracy, the system will use the previously divided test set to verify the risk assessment model. The data of the test set does not participate in the training of the model. By inputting the data of the test set into the built model, the system can predict the risk level and compare it with the actual annotated risk level. Through this verification process, the accuracy and precision of the model can be evaluated, and whether the model is overfitting or underfitting can be found. If the verification results show that the model performs poorly on certain specific abnormal types, the system will further adjust the parameters of the model, such as changing the maximum depth of the decision tree, adjusting the segmentation criteria, etc., to improve the prediction performance of the model in different risk scenarios. Finally, the risk assessment model, which has been verified and optimized many times, can effectively deal with various abnormal situations in the charging process, provide accurate risk warnings, and improve the safety and stability of the charging system.

[0086] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A battery vehicle charging integrated management system based on edge computing, characterized in that: It includes a battery vehicle end collection module, a charging pile end collection module and an intelligent management module, wherein the battery vehicle end collection module and the charging pile end collection module are respectively connected to the intelligent management module; The battery vehicle end acquisition module is used to collect battery parameters in the battery vehicle and send them to the edge computing unit, including battery model, capacity, temperature, battery voltage and charging rate; The charging pile end acquisition module is used to collect the state parameters of the charging pile load and each port of the charging pile that the power grid can support, and the state parameters of the port include output voltage, output current and load power, and send them to the edge computing unit; The intelligent management module includes an edge computing unit, a load balancing unit and a risk assessment unit, and the edge computing unit is communicatively connected with the load balancing unit and the risk assessment unit respectively; The edge computing unit is used to determine whether the current charging behavior is abnormal and determine the abnormality type according to the charging pile load, battery parameters and status parameters of each port supported by the power grid through deep learning, and distribute and send the charging pile load, battery parameters and status parameters of each port supported by the power grid to the load balancing unit or the risk assessment unit according to the abnormality type; The load balancing unit is used to adjust the load according to the reinforcement learning model based on the charging pile load that the power grid can support, the battery parameters and the status parameters of each port; The risk assessment unit is used to build a risk assessment model based on historical abnormality types, battery parameters and status parameters of each port, and output the risk level and issue an alarm prompt based on the real-time abnormality type, battery parameters and status parameters of each port through the risk assessment model.

2. According to the edge computing-based battery vehicle charging integrated management system of claim 1, it is characterized in that: The edge computing unit includes a battery central control subunit and a charging pile central control subunit, and the battery central control subunit is communicatively connected with the charging pile central control subunit; The battery central control subunit is used to receive battery parameters for standardization and send them to the charging pile central control subunit; The charging pile central control subunit is used to perform the following steps: The load of the charging pile supported by the power grid, the battery parameters and the state parameters of the port are used as the input of the feedforward neural network, and the abnormality type is output, which includes load abnormality and charging abnormality; The battery parameters and the status parameters of each port are sent to the load balancing unit or the risk assessment unit according to the abnormality type.

3. According to claim 2, the battery vehicle charging integrated management system based on edge computing is characterized in that: The training steps of the feedforward neural network include: Normalize the charging pile load, battery parameters and port status parameters that the power grid can support, and mark the abnormal types; Each parameter of the normalized power grid-supported charging pile load, battery parameters, and port status parameters is split to construct input neurons; The features of the input neurons are calculated and extracted layer by layer through the hidden layers in the feedforward neural network, and the activation function is used for nonlinear feature mapping, and the high-order features of the abnormal pattern are extracted and output through the output layer integration. The loss function is calculated based on the output results of the output layer and the marked anomaly types, and the network weights are adjusted using the back propagation algorithm.

4. According to claim 2, the battery vehicle charging integrated management system based on edge computing is characterized in that: The sending of the battery parameters and the status parameters of each port to the load balancing unit or the risk assessment unit according to the abnormality type comprises: If the abnormality type is load abnormality, the charging pile load that the power grid can support, the battery parameters and the status parameters of each port are sent to the load balancing unit; If the abnormality type is charging abnormality, the battery parameters and the status parameters of each port are sent to the risk assessment unit.

5. According to the edge computing-based battery vehicle charging integrated management system of claim 1, it is characterized in that: The load balancing unit is used to perform the following steps: The ratio of the charging pile load supported by the preset power grid to the available load of the charging pile is generated according to the real-time available load of the charging pile supported by the power grid; The current total load of the charging pile is calculated according to the state parameters of the port. The current total load of the charging pile, the real-time available load of the charging pile, and the battery parameters of each port are taken as the state, and the load adjustment of the port is taken as the action to build a reinforcement learning model, and the load adjustment is output through the reinforcement learning model.

6. The battery vehicle charging integrated management system based on edge computing according to claim 5 is characterized in that: The formula of the reinforcement learning model is as follows: Where s is the state, including the battery parameters and the state parameters of the charging pile port; a is the action, including the load regulation parameters of all ports; Q(s, a) is the expected return that can be obtained after taking action a in state s; r is the immediate reward value after performing the regulation action; γ is the discount factor; is the execution action; s′ is the next state after performing action a; a′ is the optimal regulation action; max a′ Q(s′, a′) is the maximum expected reward that can be obtained by selecting the optimal adjustment action a′ in the next state s′.

7. The battery vehicle charging integrated management system based on edge computing according to claim 6 is characterized in that: The formula of the instant reward value r is as follows: r=-α·|L current -L optimal |-β·E loss ; Among them, L current is the current load of the charging pile; L optimal E is the real-time available load of the charging pile; loss is the energy loss caused by load imbalance or power loss; α and β are the weight coefficients of load balance and energy loss respectively.

8. The battery vehicle charging integrated management system based on edge computing according to claim 1 is characterized in that: The risk assessment model is constructed based on the abnormal type, battery parameters and status parameters of each port in the history, including the following steps: Based on the abnormal data in history, collect historical data sets including abnormal type, battery parameters and status parameters of each port, and mark the corresponding risk level for each record; Divide the historical data set with completed risk level annotation into a training set and a test set; Based on the decision tree algorithm, the parameters in the training set are recursively split into two elements, and the risk level is output through the leaf nodes to obtain the risk assessment model; The risk assessment model is validated through the test set.

9. The battery vehicle charging integrated management system based on edge computing according to claim 1 is characterized in that: The risk assessment unit and the load balancing unit are both deployed on a cloud server.

Images