SOC cloud calibration method based on Ethernet protocol

Abstract

The invention discloses an SOC cloud calibration method based on an Ethernet protocol, and the method comprises the following steps: selecting a cloud server, and deploying a cloud calculation and evaluation module in the cloud server; each module of the vehicle-mounted BMS is configured based on an AUTOSAR classic platform; the vehicle-mounted BMS terminal collects performance index data of the power battery and uploads the performance index data to the cloud server; the cloud computing and evaluation module performs SOC prediction by adopting a neural network model and a Kalman filtering module to obtain an SOC predicted value S1; the cloud server sends the key parameters to the vehicle-mounted BMS end through the Ethernet; and updating an SOC calculation model in the vehicle-mounted BMS end according to the SOC predicted value S1, and obtaining an SOC true value S2 by adopting the updated SOC calculation model. The SOC cloud calibration method based on the Ethernet protocol has the advantages that efficient operation of the whole system and accurate interaction of data are ensured, and battery management of the electric vehicle is more intelligent and accurate.

Description

Technical Field

[0001] The present invention relates to a battery remaining power measurement technology, in particular to a SOC cloud calibration method based on Ethernet protocol. Background Art

[0002] Power batteries are power sources that provide power for tools, mostly referring to storage batteries that provide power for electric vehicles, electric trains, electric bicycles, and golf carts. In the field of electric transportation, the state of charge (SOC) of power batteries, as a core characterization parameter of the remaining energy of the battery, is not only a direct basis for users to evaluate the endurance ability, but also a key input for the battery management system (BMS) to achieve optimal energy distribution, charge and discharge control, and safety warning.

[0003] The BMS is the core control module of the power battery system of new energy vehicles. As the central control unit between the power energy storage unit and the vehicle energy distribution system, the BMS can jointly form three key technical systems in the field of electric vehicles with the power battery body and the vehicle control architecture. The core mechanism of the BMS is to realize the optimal management of the electrochemical energy storage system through battery state monitoring and dynamic control strategies under multi-physical field coupling conditions. For the electric vehicle battery management system, as a key state parameter representing the remaining capacity of the energy storage unit, the estimation accuracy of SOC is directly related to the reliability of the vehicle endurance mileage prediction and has become the core index for evaluating the performance of the BMS.

[0004] Currently, the electric vehicle battery management system has limitations in estimating the state of charge of the battery: due to the constraints of hardware computing, data storage capabilities, and costs, existing SOC estimation strategies usually only use historical data for a limited period. However, lithium batteries have life-like characteristics, and their structures and electrochemical parameters will gradually change over time, and their electrochemical performance will also gradually decline. As a result, traditional SOC estimation methods are difficult to accurately capture the changing characteristics of the health state of lithium batteries over time, thereby affecting the estimation accuracy.

[0005] With the rapid development of cloud computing technology and wireless communication technology, currently in the existing technology, a SOC monitoring and analysis system based on the cloud is built using the Socket interface. However, compared with the Ethernet protocol, the Socket interface has obvious deficiencies in terms of communication speed, network reliability, security, and network management and maintenance. Summary of the Invention

[0006] The present invention aims to avoid the deficiencies in the above-mentioned existing technologies and provides a SOC cloud calibration method based on Ethernet protocol to ensure the efficient operation of the entire system and the accurate interaction of data, providing a more intelligent and accurate solution for electric vehicle battery management.

[0007] The present invention adopts the following technical solutions to solve the technical problems.

[0008] The present invention provides a SOC cloud calibration method based on the Ethernet protocol. The calibration system of the SOC cloud calibration method includes a cloud platform and an in-vehicle BMS terminal; the cloud calibration method includes the following steps:

[0009] Step 1: Cloud platform construction step; select a cloud server and deploy a background management system in the cloud server; the background management system includes a cloud computing and evaluation module and a first Ethernet communication module;

[0010] Step 2: In-vehicle BMS terminal configuration step; configure the input signal acquisition and processing module, fault diagnosis module, and a second Ethernet communication module for communication with the cloud server of the in-vehicle BMS based on the AUTOSAR classic platform;

[0011] Step 3: Data acquisition and upload step; the in-vehicle BMS terminal acquires the historical key parameters of the power battery and uploads the historical key parameters to the cloud server of the cloud platform;

[0012] Step 4: Cloud computing and evaluation step; the cloud computing and evaluation module uses a neural network model and a Kalman filter module to predict SOC, and obtains the SOC prediction value S1;

[0013] Step 5: Key parameter distribution step; the cloud server sends the key parameters to the in-vehicle BMS terminal through Ethernet; the key parameters include the SOC prediction value S1;

[0014] Step 6: The in-vehicle BMS terminal receives the SOC prediction value S1, and updates the SOC calculation model in the in-vehicle BMS terminal according to the SOC prediction value S1, and obtains the SOC true value S2 by using the updated SOC calculation model.

[0015] The structural feature of a SOC cloud calibration method based on the Ethernet protocol of the present invention also lies in:

[0016] Further, in the step 1, the background management system further includes a user information management module, a battery data monitoring module, a database storage and reading module, and a device information management module.

[0017] Further, the cloud computing and evaluation module includes a neural network model and a Kalman filter module; the process of predicting the SOC prediction value S1 by using the neural network model includes the following steps:

[0018] Step 41: Data preprocessing step; preprocess the historical key parameters obtained in step 3;

[0019] Step 42: Modeling step; construct a neural network model for SOC prediction and train the neural network model according to historical key parameters;

[0020] Step 43: Filtering step; initialize the Kalman filter module and establish a state space model of the battery; the state space model includes a state equation and an observation equation;

[0021] Step 44: Prediction step; input the latest performance index data collected in real time on site into the trained neural network model to obtain the SOC prediction value S1;

[0022] Step 45: Output step; output the SOC prediction value S1.

[0023] Further, in the step 41, the preprocessing of the historical key parameters includes data cleaning and data normalization.

[0024] Further, in the step 42, the process of constructing the neural network model includes the following steps:

[0025] Step 421: Select a multi-layer perceptron MLP as the architecture of the neural network model;

[0026] Step 422: Set the layer configuration and parameters of the multi-layer perceptron MLP;

[0027] Step 423: Determine the training strategy of the multi-layer perceptron MLP;

[0028] Step 424: Use historical key parameters to train the multi-layer perceptron MLP.

[0029] Further, in the step 421, the multi-layer perceptron MLP includes an input layer, a hidden layer, and an output layer.

[0030] Further, in the step 43, the state equation is used to describe the dynamic change of the battery SOC; the observation equation is used to describe the relationship between the measurable performance parameters of the battery and the SOC.

[0031] Further, in the step 6, the in-vehicle BMS updates the SOC calculation model inside the BMS according to the key parameters received from the cloud platform, and calculates the true SOC value S2 according to the updated SOC calculation model.

[0032] The present invention also provides an electronic device, including at least one processor and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the SOC cloud calibration method based on the Ethernet protocol.

[0033] The present invention also provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the SOC cloud calibration method based on the Ethernet protocol as described above.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0035] The present invention discloses a SOC cloud calibration method based on the Ethernet protocol, which includes the following steps: Select a cloud server and deploy a cloud computing and evaluation module in the cloud server; Configure each module of the in-vehicle BMS based on the AUTOSAR classic platform; The in-vehicle BMS collects performance index data of the power battery and uploads the performance index data to the cloud server; The cloud computing and evaluation module uses a neural network model and a Kalman filter module to perform SOC prediction to obtain a SOC prediction value S1; The cloud server sends key parameters to the in-vehicle BMS via Ethernet; Update the SOC calculation model in the in-vehicle BMS according to the SOC prediction value S1, and use the updated SOC calculation model to obtain the true SOC value S2.

[0036] The SOC cloud calibration method based on the Ethernet protocol of the present invention has the advantages of ensuring the efficient operation of the entire system and the accurate interaction of data, making the battery management of electric vehicles more intelligent and precise, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a system block diagram of a SOC cloud calibration method based on the Ethernet protocol of the present invention.

[0038] Figure 2 It is a block diagram of a background management system based on a cloud platform for a SOC cloud calibration method based on the Ethernet protocol of the present invention.

[0039] Figure 3 It is a BMS system block diagram of a SOC cloud calibration method based on the Ethernet protocol of the present invention.

[0040] The following will further illustrate the present invention through specific embodiments in conjunction with the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] See Figures 1 to 3 , the present invention provides a SOC cloud calibration method based on the Ethernet protocol. The calibration system of the SOC cloud calibration method includes a cloud platform and an in-vehicle BMS terminal; The cloud calibration method includes the following steps:

[0042] Step 1: Cloud platform construction step; Select a cloud server and deploy a background management system in the cloud server; The background management system includes a cloud computing and evaluation module and a first Ethernet communication module;

[0043] Step 2: Configuration steps at the in-vehicle BMS side; Configure the input signal acquisition and processing module, fault diagnosis module, and the second Ethernet communication module for communication with the cloud server on the in-vehicle BMS based on the AUTOSAR classic platform;

[0044] Such as Figure 3 is the framework diagram of the in-vehicle BMS side of the present invention.

[0045] Step 3: Data acquisition and upload steps; The in-vehicle BMS side acquires the historical key parameters of the power battery and uploads the historical key parameters to the cloud server of the cloud platform;

[0046] The in-vehicle BMS acquires the historical key parameters of the battery system in real time, including the terminal voltage U of the single battery cell , total current I, battery temperature Th, etc. Preprocess the acquired signals, such as filtering, amplification, analog-to-digital conversion, etc. Package the preprocessed data into data frames conforming to the Ethernet protocol and upload them to the cloud computing center through the Ethernet communication module. In addition, the SOC will be preliminarily estimated by the ampere-hour integration method at the in-vehicle side.

[0047] Step 4: Cloud computing and evaluation steps; The cloud computing and evaluation module uses a neural network model and a Kalman filter module for SOC prediction to obtain the SOC predicted value S1;

[0048] The cloud computing center receives the historical key parameters uploaded by the in-vehicle BMS and stores them in the database. Regularly analyze the historical key parameters updated to the current moment, identify and update the battery parameters, such as internal resistance, maximum discharge capacity, etc., and perform SOC prediction through neural network and Kalman filter algorithms to obtain the SOC predicted value S1. The specific prediction steps are shown in Process 2.

[0049] Step 5: Key parameter distribution steps; The cloud server sends the key parameters to the in-vehicle BMS side through Ethernet; The key parameters include the SOC predicted value S1;

[0050] The cloud computing center regularly distributes the calculated key parameters, such as maximum discharge capacity, SOH, etc., to the corresponding in-vehicle BMS through the Ethernet protocol. After receiving these parameters, the in-vehicle BMS will recalculate or calibrate the SOC through XcpOnEth, that is, the calibration protocol stack based on Ethernet.

[0051] Step 6: The in-vehicle BMS side receives the SOC predicted value S1, and updates the SOC calculation model in the in-vehicle BMS side according to the SOC predicted value S1, and obtains the SOC true value S2 by using the updated SOC calculation model.

[0052] Such as Figure 1This is a block diagram of the SOC cloud calibration system of the present invention, which mainly includes two major parts: a cloud platform and an in-vehicle BMS terminal. As Figure 2 In the cloud platform, a neural network model and a Kalman filter module are used for prediction to obtain the SOC predicted value S1.

[0053] As Figure 3 , the in-vehicle BMS terminal constructs a two-way communication link with the cloud platform through Ethernet to improve data transmission efficiency and security. In order to ensure the efficient operation of the BMS system and the accurate interaction of data, the present invention adopts the AUTOSAR classic platform, which is an automotive development system architecture and includes multiple standardized protocol stacks such as CAN communication, Ethernet communication, calibration protocol stack, and diagnostic protocol stack. Through the AUTOSAR classic platform, functions such as input signal acquisition and processing, fault diagnosis, Ethernet communication, battery charge and discharge power estimation, thermal management control, relay control, and SOC estimation and calibration of the BMS can be developed in a standardized and modular manner, greatly improving the software reusability and portability, and solving problems such as high software and hardware coupling degree, poor portability, and long development cycle in the software development of traditional battery management systems. The BMS software design architecture adopting the AUTOSAR architecture is as Figure 3 .

[0054] In specific implementation, in step 1, the background management system further includes a user information management module, a battery data monitoring module, a database storage and reading module, and a device information management module.

[0055] As Figure 2 This is a functional block diagram of the background management system of the cloud platform. The background management system is an important link for the battery management system to implement data management. The user information management module is mainly used for user account registration, information management, permission setting, password management, etc. The battery data monitoring module is used for real-time monitoring of various data of the battery system, such as the voltage of single cells, total current, total voltage, etc. The device information management module is mainly used for managing the name, product ID, APIKey, status, location, running time, and device introduction of the battery management system.

[0056] The cloud computing and evaluation module is used for analyzing the historical key parameters at the current moment. The historical key parameters are shown in Table 1.

[0057] A high-performance server is selected as the cloud server, and a cloud computing and evaluation module is set therein. In the database configured by the cloud server, the historical key parameters of the battery are stored. The historical key parameters include, but are not limited to, the terminal voltage U of the single cell cell, information such as the total current I, battery temperature Th, etc. The database of the cloud server sets up a data backup and recovery mechanism to ensure the security and reliability of the data. The historical key parameters are shown in Table 1 below, where T is the length of the time series.

[0058] Table 1: Historical Key Parameters of the Cloud Server Database

[0059]

[0060] In specific implementation, the cloud computing and evaluation module includes a neural network model and a Kalman filter module; the process of predicting the SOC prediction value S1 using the neural network model includes the following steps:

[0061] Step 41: Data preprocessing step; preprocess the historical key parameters obtained in step 3;

[0062] Step 42: Modeling step; construct a neural network model for SOC prediction, and train the neural network model according to the historical key parameters;

[0063] Step 43: Filtering step; initialize the Kalman filter module and establish a state space model of the battery; the state space model includes a state equation and an observation equation;

[0064] Step 44: Prediction step; input the latest performance index data collected on-site in real time into the trained neural network model to obtain the SOC prediction value S1;

[0065] Step 45: Output step; output the SOC prediction value S1.

[0066] In specific implementation, in step 41, the preprocessing of the historical key parameters includes data cleaning and data normalization.

[0067] Perform data cleaning operations on the collected historical key parameters to process outliers, missing values, etc. of each item of data. Then perform data normalization processing. To accelerate the convergence speed of the neural network, the minimum-maximum normalization method is used to normalize the data. The formula for the normalization is shown in the following formula (1).

[0068]

[0069] In formula (1), x is the original data of a certain historical key parameter, x min and x max are the minimum and maximum values of this data, and x norm is the normalized data. The historical key parameters include the terminal voltage U cell , charge and discharge current I, battery temperature Th, etc. In specific implementation, the BMS is used to collect the performance indicators of the battery at 1000HZ Continuously collect data on performance metrics for the sampling frequency.

[0070] In specific implementation, in step 42, the process of constructing the neural network model includes the following steps:

[0071] Step 421: Select the Multilayer Perceptron (MLP) as the architecture of the neural network model;

[0072] Step 422: Set the layer configuration and parameters of the Multilayer Perceptron (MLP);

[0073] Step 423: Determine the training strategy of the Multilayer Perceptron (MLP);

[0074] Step 424: Train the Multilayer Perceptron (MLP) using historical key parameters.

[0075] The present invention uses the Multilayer Perceptron (MLP) as the neural network architecture, which has a simple structure and is suitable for processing static data; the overall design of the neural network architecture includes the MLP network structure design and the training optimization strategy.

[0076] In specific implementation, in step 421, the Multilayer Perceptron (MLP) includes an input layer, a hidden layer, and an output layer.

[0077] The input layer inputs the previously collected and preprocessed historical key parameters. The historical key parameters are used as input data. In specific implementation, a column vector a 0 = [U cell , I, Th, SOC avg T can be used as the static features input after preprocessing. a 0 represents the activation value of the input layer. a 0 is used to transfer the original features to the first hidden layer and initiate the forward propagation process. The hidden layer of the present invention includes l layers, and the calculation formula of the l-th hidden layer is shown in the following formula (2).

[0078] Z (l) = W (l) * a (l-1) + b (l)

[0079] a (l) = σ(z (l) ) l = (natural numbers 1, 2, 3......) (2)

[0080] In formula (2), W (l) represents the weight matrix of the l-th hidden layer, which is used to perform weighted summation of the features of this hidden layer, as shown in formula (4); b (l)Indicates that the bias vector is used to adjust the output value Z of the l-th hidden layer (l) The reference offset, and the calculation can refer to formula (5); σ(·) is the activation function a (l) Indicates the activation value of the l-th hidden layer. In one embodiment, the number of hidden layers l = 2, as shown in Table 2 below

[0081] The output layer is used to output the final SOC prediction value S1, and the calculation formula for the output is shown in formula (3) below

[0082] S1 = σ[W (out) * l +b (out) (3)

[0083] In formula (3), W (out) Indicates the weight matrix of the output layer, which is used to perform weighted summation of the features of the output layer, as shown in formula (4); b (out) Indicates the bias vector of the output layer, which is used to adjust the reference offset of the output value S1 of the output layer, as shown in formula (5); Z l Indicates the output value of the last hidden layer; σ(·) is the activation function, specifically as shown in formula (2)

[0084] In step 422, according to the SOC task characteristics in the BMS, the specific hierarchical configuration and parameter settings are shown in Table 2

[0085] Table 2: Hierarchical configuration and parameters

[0086]

[0087] The details of parameter initialization specifically include He initialization and output layer bias. Among them, the purpose of He initialization is to adapt to the characteristics of the ReLU activation function and alleviate the problem of gradient disappearance or explosion, as shown in formula (4) below

[0088]

[0089] In formula (4), Indicates the weight value at the m-th row and n-th column in the weight matrix W (l) of the l-th hidden layer, d l-1 Indicates the number of nodes in the l-1 layer, Indicates the bias term of the t-th neuron in the l-th hidden layer

[0090] The initialization of the hidden layer and output layer bias is used to avoid introducing artificial biases during the initialization stage, allowing the model to autonomously learn the bias values through training

[0091] The mathematical expression for the initialization of the output layer bias is shown in formula (5) below

[0092]

[0093] In formula (5), where l is the number of hidden layers, b (l+1) represents the bias value of the output layer, and n train is the number of training set samples. y h is the SOC sample value of the h-th sample in the training set; y h is derived from historical key parameters.

[0094] For the training of the neural network model, parameters of a large number of batteries under different working conditions, temperatures, and aging states are collected as training samples. A neural network model is constructed using a deep learning framework, and the non-linear relationship between battery characteristics and SOC is learned through training. The Kalman filter algorithm mainly combines the original battery model to online identify and update battery parameters such as internal resistance and maximum discharge capacity.

[0095] In step 423, the training optimization strategy mainly includes the following two steps: constructing a loss function and adaptively adjusting the learning rate.

[0096] (1) Battery data usually has noise, such as problems like voltage or current jitter. The stage of constructing the loss function is mainly used to suppress the excessive sensitivity of the neural network model to sensor noise, such as the instantaneous fluctuations in current sampling. In the present invention, MSE is used as the optimization objective to construct the loss function L as shown in formula (6) below.

[0097]

[0098] In formula (6), where W (l) represents the weight matrix of the l-th hidden layer, ∥·∥F is the Frobenius norm, which represents the constraint on the weight amplitude; n batch is the batch size, which represents the number of training set samples input into the model at one time, and y i represents the SOC sample value of the i-th sample, and y i ′ represents the model prediction value S1 of the i-th sample, and λ is the strength of controlling the regularization term, which is used to prevent the model from overfitting.

[0099] (2) The adaptive process of the learning rate α includes a cosine annealing scheduling process and a regularization process. The loss function for battery SOC estimation may have multiple local minima, such as the capacity decay patterns at different temperatures. Periodically resetting the learning rate α is to help jump out of local traps.

[0100] ① The formula for cosine annealing scheduling is shown in formula (7) below.

[0101]

[0102] In formula (7), α max= 0.01, which is used to define the peak value of the learning rate α within one cycle. A relatively high learning rate is adopted in the initial stage to help the model quickly escape from local optima or flat regions. α min = 0.0001, which is used to define the lowest value of the learning rate within one cycle. A relatively small learning rate is adopted in the later stage to enable the model to finely search for the optimal solution at the bottom of the loss function. Cycle T cycle = 50 epochs, which is used to control the cycle during which the learning rate α decreases from α max to α min .

[0103] During the regularization process, it is first necessary to go through the dropout stage, which is only activated during training to reduce the sensitivity to data noise characteristics. The process of the dropout stage is shown in Equation (8). Among them, the packet loss rate p = 0.2, indicating that the activation value a z (l) will be set to 0 with probability p, and a z (l) represents the activation value of the z-th neuron in the l-th hidden layer.

[0104]

[0105] During the dropout stage, multiple forward propagations are performed on the same input, and some neurons will be randomly deactivated each time. Suppose a total of N forward propagations are performed, and the activation values of the z-th neuron in the l-th hidden layer for the 1st to N-th forward propagations are a z (l,1) , z (l,2) ,…, z (l,N) , where j represents the j-th forward propagation. Then, the activation value variance and confidence are obtained from Equation (9), where represents the activation value variance, and Con represents the confidence.

[0106]

[0107] The early stopping rule is to terminate the training when the validation set loss has not decreased for 10 consecutive epochs. Its role is, on the one hand, to limit the amount of data. When the cost of battery aging experiments is high and the scale of training data is small, early stopping can prevent overfitting on small datasets. On the other hand, when the number of model parameters selected by early stopping is small, it is more suitable for edge deployment.

[0108] Specifically, in step 43, the state equation is used to describe the dynamic change of the battery SOC; the observation equation is used to describe the relationship between the measurable battery performance parameters and the SOC.

[0109] The state - space model of the battery includes a state equation and an observation equation. The state equation describes the dynamic change of the battery's SOC, and the observation equation describes the relationship between the measurable battery parameters (such as the terminal voltage U cell ) and the SOC.

[0110] The state equation is shown in the following formula (10).

[0111] SOC k+1 = A * SOC k + B * u k + w k (10)

[0112] The observation equation is shown in the following formula (11).

[0113] U cellk = H * SOC k + v k (11)

[0114] In formulas (10) and (11), SOC k is the SOC state value at time k, A is the state - transition matrix, B is the output matrix, u k is the input variable, w k is the process noise, U cellk is the observed terminal - voltage value at time k, H is the observation matrix, v k is the observation noise; the initial SOC state is set as SOC0 and the initial covariance matrix is P0.

[0115] The initial SOC0 is estimated by the open - circuit voltage method. Online SOC estimation: The historical key parameters collected and pre - processed in real time are input into the trained neural - network model to obtain the preliminary predicted value of SOC, SOC start ; Kalman filtering update is performed, which specifically includes time update and observation update. The specific update process is as follows.

[0116] (1) The time update is to predict the SOC state according to the state equation and update the covariance matrix P at the same time. The specific formula is shown in the following formula (12).

[0117] SOc k = A * SOC k-1 + B * u k

[0118] P k = A * P k-1 + A T + Q (12)

[0119] In formula (11), SOC k is the SOC state value at time k, Pk is the predicted covariance matrix at time k, and Q is the covariance matrix of the process noise. A T is the transpose matrix of the state transition matrix A.

[0120] (2) The measurement update combines the neural network predicted value SOC start and SOC k to update the SOC state. The specific formula is shown in the following formula (13).

[0121]

[0122] SOC k = SOC k-1 + K k *(z k - H * y k-1 )

[0123] P k = (I - K k * H) * P k-1 (13)

[0124] In formula (13), K k is the Kalman gain, R is the covariance matrix of the observation noise, SOC k is the updated SOC state value at time k, and P k is the updated covariance matrix at time k. H T is the transpose matrix of the observation matrix.

[0125] Take the SOC state SOC k updated by the Kalman filter as the final SOC predicted value S1 for output, and send it to the vehicle-mounted BMS through the Ethernet protocol. Continuously monitor the accuracy and stability of the SOC estimation result. If it is found that the estimation error is large, the neural network model can be retrained or the parameters of the Kalman filter can be adjusted.

[0126] Specifically, in step 6, the vehicle-mounted BMS updates the SOC calculation model inside the BMS according to the key parameters received from the cloud platform, and calculates the true value S2 of SOC according to the updated SOC calculation model.

[0127] During the update process, verify the improvement effect of the SOC estimation accuracy by comparing with the actual measurement data.

[0128] Step 61: The vehicle-mounted BMS parses the parameter packet of the key parameters sent by the cloud platform through the XcpOnEth protocol and checks the integrity of the key parameters; the key parameters include parameters such as the SOC predicted value S1 calculated by the neural network model and the confidence level Zx;

[0129] Step 62: Determine the final true value SOC2 based on the confidence value Zx, the SOC prediction value S1 of the cloud, and the estimated value SOC of the vehicle-mounted terminal l , and determine the final true value SOC2;

[0130] Step 621: If the cloud SOC confidence Zx is greater than 0.9, i.e., Zx > 0.9, the measure taken is to directly use the cloud SOC prediction value S1 to overwrite the estimated value SOC of the vehicle-mounted terminal l ; and use the overwritten SOC l value as the SOC true value S2 and include it in the key parameters for uploading;

[0131] Step 622: If the cloud SOC confidence is between 0.6 and 0.9, then SOC is updated using formula (14).

[0132]

[0133] In formula (14), is the confidence issued by the cloud, β is the local SOC weight, and is i.e., SOC l is the SOC value calculated by the vehicle-mounted terminal; and use the calculated SOC2 as the SOC true value S2 and include it in the key parameters for uploading;

[0134] Step 623: If the cloud SOC confidence is less than 0.6, or the cloud SOC prediction value S1 is more than 15% greater than the SOC estimated value of the vehicle-mounted terminal, then trigger an SOC alarm anomaly, suspend the update, and report to the cloud to request manual verification.

[0135] The present invention also provides an electronic device, including at least one processor and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above-mentioned SOC cloud calibration method based on the Ethernet protocol.

[0136] The present invention also provides a non-transitory computer-readable storage medium storing computer instructions, and the computer instructions are used to cause the computer to execute the SOC cloud calibration method based on the Ethernet protocol.

[0137] In specific implementation, Ethernet communication can ensure the security and efficiency of information transmission. Compared with the traditional cloud calibration method, the cloud calibration based on the Ethernet protocol has the following advantages, as shown in Table 3 specifically.

[0138] Table 3 Advantages of the Ethernet communication solution

[0139] Pain point Socket communication solution Ethernet communication solution Security Rely on custom encryption and is easily cracked Standardized encryption Real-time performance The delay is uncontrollable and it is difficult to meet the TSN requirements TSN supports deterministic delay Bandwidth The bandwidth is limited and it is difficult to transmit large data 100 Mbps / 1 Gbps bandwidth, supporting high-definition data transmission Standardization The protocols are fragmented and the compatibility is poor Complies with ISO / IEEE standards and is cross-platform compatible

[0140] A SOC cloud estimation method based on the Ethernet protocol of the present invention calculates the SOC estimation in the cloud, uses Ethernet communication instead of Socket communication for communication between the cloud and the BMS side, and designs the BMS architecture based on the AUTOSAR platform, having the following technical characteristics.

[0141] 1. Breaking through the limitations of traditional SOC estimation methods: Aiming at the problem that the existing vehicle-mounted BMS is limited by hardware computing power and cost and can only use limited historical data for SOC estimation, a solution based on cloud computing is proposed. By uploading long-term historical data to the cloud for analysis and combining with the dynamic modeling of the aging characteristics of lithium batteries, accurate capture of key parameters such as the state of health (SOH) and internal resistance of the battery is achieved, thereby improving the SOC estimation accuracy and enhancing the reliability of the vehicle's cruising range prediction;

[0142] 2. Optimization of communication efficiency and security: The Ethernet protocol is used to replace the Socket interface, significantly improving the data transmission speed, reducing latency, and meeting the real-time requirements. In addition, the Ethernet protocol has a more perfect network management mechanism and security protection system to ensure the stability and security of data transmission;

[0143] Aiming at the deficiencies of the Socket interface in terms of communication speed, network reliability, security, and maintainability, the Ethernet protocol is used to build a two-way communication link between the vehicle-mounted BMS and the cloud computing center. Through standardized network protocols, the data transmission efficiency and security are improved, and the bottlenecks of traditional Socket communication in real-time and stability are solved, providing more reliable technical support for cloud data interaction.

[0144] 3. System intelligent and configurable management: Based on the configurable design of the AUTOSAR classic platform, automatic parsing and transmission of messages are realized, reducing the manual coding workload, improving the development efficiency and reducing the maintenance cost. In addition, the configurable architecture enhances the system scalability, facilitating future function upgrades (such as adding new sensors or algorithm modules) without large-scale reconstruction of the underlying code.

[0145] Based on the architecture advantages of the AUTOSAR classic platform, automatic parsing and transmission of messages are realized through configurable management, simplifying the system development and maintenance process. At the same time, key parameters of the cloud computing center (such as the maximum discharge capacity, SOH, etc.) are periodically sent to the vehicle-mounted BMS to support the dynamic update of the SOC calculation model, thereby constructing an intelligent battery management system with self-optimization capabilities.

[0146] A SOC cloud estimation method based on the Ethernet protocol of the present invention has the following basic working mode: The in-vehicle BMS collects the basic information of the battery system and uploads it to the cloud computing center through the Ethernet protocol. The cloud computing center stores the data and analyzes and calculates the historical data updated to the current moment according to a certain model. Then, key data such as the maximum discharge capacity, internal resistance, state of health (SOH), etc., which cannot be analyzed and calculated on the in-vehicle BMS, are periodically downloaded to the corresponding in-vehicle BMS through the Ethernet protocol. The in-vehicle BMS recalculates or calibrates the SOC based on these data. In addition, based on the AUTOSAR classic platform, the upload and download of messages are only passed to the corresponding protocol stack for parsing through configuration, thus ensuring the efficient operation of the entire system and the accurate interaction of data, providing a more intelligent and accurate solution for electric vehicle battery management.

[0147] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claims involved.

[0148] In addition, it should be understood that although this specification is described according to embodiments, not every embodiment only contains an independent technical solution. This narrative way of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A SOC cloud calibration method based on the Ethernet protocol, characterized in that The calibration system includes a cloud platform and an in-vehicle BMS terminal; the cloud calibration method includes the following steps: Step 1: Cloud platform construction step; Select a cloud server and deploy a background management system in the cloud server; The background management system includes a cloud computing and evaluation module and a first Ethernet communication module; Step 2: In-vehicle BMS terminal configuration step; Based on the AUTOSAR classic platform, configure the input signal acquisition and processing module, fault diagnosis module, and a second Ethernet communication module for communication with the cloud server of the in-vehicle BMS; Step 3: Data acquisition and upload step; The in-vehicle BMS terminal collects the historical key parameters of the power battery and uploads the historical key parameters to the cloud server of the cloud platform; Step 4: Cloud computing and evaluation step; The cloud computing and evaluation module uses a neural network model and a Kalman filter module to predict the SOC and obtains the SOC prediction value S1; Step 5: Key parameter distribution step; The cloud server sends the key parameters to the in-vehicle BMS terminal through Ethernet; The key parameters include the SOC prediction value S1; Step 6: The in-vehicle BMS terminal receives the SOC prediction value S1, and updates the SOC calculation model in the in-vehicle BMS terminal according to the SOC prediction value S1, and obtains the SOC true value S2 by using the updated SOC calculation model.

2. The SOC cloud calibration method based on the Ethernet protocol according to claim 1 is characterized in that, In the step 1, the background management system further includes a user information management module, a battery data monitoring module, a database storage and reading module, and a device information management module.

3. A SOC cloud calibration method based on the Ethernet protocol according to claim 1, characterized in that, The cloud computing and evaluation module includes a neural network model and a Kalman filter module; The process of using the neural network model to predict the SOC prediction value S1 includes the following steps: Step 41: Data preprocessing step; Preprocess the historical key parameters obtained in step 3; Step 42: Modeling step; Construct a neural network model for SOC prediction, and train the neural network model according to the historical key parameters; Step 43: Filtering step; Initialize the Kalman filter module and establish a state space model of the battery; The state space model includes a state equation and an observation equation; Step 44: Prediction step; Input the latest performance index data collected in real time on-site into the trained neural network model to obtain the SOC prediction value S1; Step 45: Output step; Output the SOC prediction value S1.

4. The SOC cloud calibration method based on Ethernet protocol according to claim 3, characterized in that, In the step 41, the preprocessing of the historical key parameters includes data cleaning and data normalization.

5. The SOC cloud calibration method based on Ethernet protocol according to claim 3, characterized in that In the step 42, the process of constructing the neural network model includes the following steps: Step 421: Select the multi-layer perceptron MLP as the architecture of the neural network model; Step 422: Set the layer configuration and parameters of the multi-layer perceptron MLP; Step 423: Determine the training strategy of the multi-layer perceptron MLP; Step 424: Use the historical key parameters to train the multi-layer perceptron MLP.

6. The SOC cloud calibration method based on Ethernet protocol according to claim 5, characterized in that, In the step 421, the multi-layer perceptron MLP includes an input layer, a hidden layer, and an output layer.

7. The SOC cloud calibration method based on the Ethernet protocol according to claim 3, characterized in that, In the step 43, the state equation is used to describe the dynamic change of the battery SOC; The observation equation is used to describe the relationship between the measurable battery performance parameters and the SOC.

8. A SOC cloud calibration method based on the Ethernet protocol according to claim 1, characterized in that, In step 6, the in-vehicle BMS updates the SOC calculation model inside the BMS according to the key parameters received from the cloud platform, and calculates the true value S2 of the SOC according to the updated SOC calculation model.

9. An electronic device, comprising at least one processor and a memory communicatively connected to the at least one processor; characterized in that, The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the Ethernet protocol-based SOC cloud calibration method according to any one of claims 1-8.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are used to cause the computer to execute the Ethernet protocol-based SOC cloud calibration method according to any one of claims 1-8.

Images