Battery SOC correction method and system, cloud estimation method and device and vehicle

Abstract

The invention relates to a battery SOC correction method and system, a cloud estimation method and device and a vehicle. Real-time and accurate estimation of the battery SOC is achieved. The method comprises the following steps: acquiring real-time operation parameters of a battery; inputting the real-time temperature, the real-time voltage and the real-time current of the battery into a pre-constructed lightweight estimation model to obtain a first estimation value; carrying out ampere-hour integration on the real-time current of the battery to obtain a second estimated value; obtaining an SOC initial value according to the first estimation value and the second estimation value; when the operation condition of the battery belongs to the predetermined key condition, uploading the SOC initial value and a data segment containing the real-time operation parameters of the battery to a cloud end, so that the cloud end determines whether to carry out lightweight estimation network updating and SOC correction value issuing; and when the updated lightweight estimation network and the SOC correction value issued by the cloud are received, updating the lightweight estimation network, and obtaining a final SOC value of the battery according to the SOC initial value and the SOC correction value.

Description

Technical Field

[0001] The present application relates to the field of batteries, and specifically to a battery SOC correction method and system, a cloud-based estimation method, device, and vehicle. Background Art

[0002] Lithium iron phosphate batteries (LiFePO4, LFP for short) have gradually become a key battery technology in the field of electric vehicles and energy storage due to their cost-effectiveness, high safety and long cycle life.

[0003] In the prior art, battery SOC estimation is typically based on the ampere-hour integration method. However, this method is unsuitable for LFP battery SOC estimation. The ampere-hour integration method relies on the precise measurement and time integration of the battery's charge and discharge currents. However, the voltage platform of LFP batteries is too flat, making it impossible to regularly correct the integration result through distinct voltage inflection points like ternary batteries. As a result, current measurement errors, battery self-discharge, and complex electrochemical processes lead to the accumulation of integration errors, causing the SOC estimate obtained by the ampere-hour integration method to deviate from the actual value. In addition, the electrochemical performance of lithium iron phosphate batteries varies significantly under different operating conditions such as temperature and charge and discharge rates. The traditional single ampere-hour integration estimation model cannot maintain high-precision SOC estimation under all operating conditions.

[0004] To this end, existing technologies also include calibrating the SOC estimated using the ampere-hour integration method by utilizing cloud-based big data models. Theoretically, utilizing cloud-based big data for SOC calibration can improve the accuracy of SOC estimation. However, existing methods typically require uploading all data from the battery's operation process to a cloud server. This not only places extremely high demands on the communication network bandwidth, but can also cause data transmission delays or even interruptions, impacting the timeliness and accuracy of calibration. This results in an inability to respond to rapid dynamic changes in battery status in real time, making it difficult to meet the demand for accurate real-time estimation of battery SOC during electric vehicle driving. Furthermore, vehicle-side devices (such as onboard BMSs) have limited computing resources due to cost, power consumption, and space constraints. Complex, high-precision big data models deployed in the cloud are difficult to deploy directly on these resource-constrained edge devices. Summary of the Invention

[0005] The present application provides a battery SOC correction method and system, a cloud-based estimation method, device and vehicle, to achieve real-time and accurate estimation of battery SOC.

[0006] The technical solution of this application is: This application provides a cloud-based battery SOC estimation method, including: Obtain the initial SOC value and the data segment containing the real-time operating parameters of the battery uploaded by the vehicle; Running a first estimation model based on a data segment containing real-time operating parameters of the battery to obtain a third SOC estimation value; Running a second estimation model based on a data segment containing real-time operating parameters of the battery to obtain a fourth SOC estimation value; Obtaining an SOC calibration value according to the third SOC estimated value and the fourth SOC estimated value; determining whether to update a lightweight estimation model according to the SOC calibration value and the SOC initial value; When it is determined that the lightweight estimation model needs to be updated, the lightweight estimation model is updated using the SOC calibration value and the data segment containing the real-time operating parameters of the battery; The updated lightweight estimation network is sent to the vehicle side, and the difference between the SOC calibration value and the SOC initial value is sent to the vehicle side as the SOC correction value.

[0007] Preferably, the step of obtaining the SOC calibration value according to the third SOC estimated value and the fourth SOC estimated value comprises: multiplying the third SOC estimated value by a third preset weight coefficient to obtain a third product value; multiplying the fourth SOC estimated value by a fourth preset weight coefficient to obtain a fourth product value; Adding the third product value and the fourth product value to obtain an SOC calibration value; The sum of the third preset weight coefficient and the fourth preset weight coefficient is 1; The third preset weight coefficient and the fourth preset weight coefficient are determined according to the aging degree of the battery and / or the current working condition of the battery.

[0008] This application also provides a method for correcting battery SOC, including: Acquire real-time operating parameters of the battery; the real-time operating parameters of the battery include: real-time temperature, real-time voltage and real-time current of the battery; Inputting the real-time temperature, real-time voltage, and real-time current of the battery into a pre-built lightweight estimation model to obtain a first SOC estimation value; Integrate the real-time current of the battery in ampere hours to obtain a second SOC estimation value; Obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value; When it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, the initial SOC value and a data segment containing the real-time operating parameters of the battery are uploaded to the cloud for the cloud to determine whether to perform a lightweight estimation network update and send a corrected SOC value; Upon receiving the updated lightweight estimation network and SOC correction value sent from the cloud, updating the lightweight estimation network and obtaining the final SOC value of the battery based on the initial SOC value and the SOC correction value; When the updated lightweight estimation network and SOC correction value sent by the cloud are not received, the initial SOC value is used as the final SOC value of the battery; Among them, the cloud uses the above-mentioned cloud-based estimation method of the battery SOC to determine the updated lightweight estimation network and SOC correction value.

[0009] Preferably, if the operating condition of the battery satisfies at least one of the following conditions: The rate of change of the battery voltage exceeds a first predetermined rate of change; The battery temperature change rate exceeds a second predetermined change rate; The battery switches between charge and discharge states; Determine that the battery's operating conditions fall within predetermined critical conditions.

[0010] Preferably, the step of obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value includes: multiplying the first SOC estimation value by a first preset weight coefficient to obtain a first product value; multiplying the second SOC estimated value by a second preset weight coefficient to obtain a second product value; Adding the first product value and the second product value to obtain an initial SOC value; The sum of the first preset weight coefficient and the second preset weight coefficient is 1; The first preset weight coefficient and the second preset weight coefficient are determined according to the real-time temperature of the battery.

[0011] Preferably, the step of obtaining the final SOC value of the battery according to the initial SOC value and the corrected SOC value includes: The initial SOC value and the corrected SOC value are added to obtain a final SOC value of the battery.

[0012] Preferably, when it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, uploading the initial SOC value and the data segment containing the real-time operating parameters of the battery to the cloud includes: When the SOC of the battery is greater than or equal to a preset SOC and the discharge rate of the battery is less than or equal to a preset discharge rate, uploading the data segment to the cloud at a first preset frequency; When the SOC of the battery is less than a preset SOC or the discharge rate of the battery is greater than a preset discharge rate, uploading the data segment to the cloud at a second preset frequency; The first preset frequency is lower than the second preset frequency.

[0013] The present application also provides a battery SOC correction system, comprising: a controller and a vehicle-mounted network terminal connected thereto; the controller is configured to: Acquire real-time operating parameters of the battery; the real-time operating parameters of the battery include: real-time temperature, real-time voltage and real-time current of the battery; Inputting the real-time temperature, real-time voltage, and real-time current of the battery into a pre-built lightweight estimation model to obtain a first SOC estimation value; Integrate the real-time current of the battery in ampere hours to obtain a second SOC estimation value; Obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value; When it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, the initial SOC value and a data segment containing the real-time operating parameters of the battery are uploaded to the cloud via the vehicle-mounted network terminal, so that the cloud can determine whether to perform a lightweight estimation network update and send a corrected SOC value; When the updated lightweight estimation network and SOC correction value sent from the cloud are received through the vehicle-mounted network terminal, the lightweight estimation network is updated, and the final SOC value of the battery is obtained based on the initial SOC value and the SOC correction value; When the updated lightweight estimation network and SOC correction value sent from the cloud are not received through the vehicle-mounted network terminal, the initial SOC value is used as the final SOC value of the battery; Among them, the cloud uses the above-mentioned cloud-based estimation method of the battery SOC to determine the updated lightweight estimation network and SOC correction value.

[0014] This application also provides a cloud-based battery SOC estimation device, including: The acquisition module is used to obtain the initial SOC value uploaded by the vehicle and the data segment containing the real-time operating parameters of the battery; a first estimation module, configured to execute a first estimation model based on a data segment containing real-time operating parameters of the battery to obtain a third SOC estimation value; a second estimation module, configured to run a second estimation model based on a data segment containing real-time operating parameters of the battery to obtain a fourth SOC estimation value; an SOC calibration value estimating module, configured to obtain an SOC calibration value according to the third SOC estimation value and the fourth SOC estimation value; A model update judgment module, configured to determine whether to update the vehicle-side lightweight estimation model based on the SOC calibration value and the SOC initial value; a model updating module, configured to update the lightweight estimation model on the vehicle side using the SOC calibration value and a data segment containing real-time operating parameters of the battery when determining that the lightweight estimation model on the vehicle side needs to be updated; The sending module is used to send the updated lightweight estimation network to the vehicle side, and send the difference between the SOC calibration value and the SOC initial value as the SOC correction value to the vehicle side.

[0015] The present application also provides a vehicle, comprising the above-mentioned battery SOC correction system.

[0016] The beneficial effects of the present invention are: Due to the limited computing power of the vehicle-side controller, a lightweight estimation model is deployed on the vehicle. This model and ampere-hour integration are used to estimate the SOC under normal operating conditions, enabling real-time SOC estimation. For certain critical operating conditions, data segments at relevant time points are uploaded to the cloud, and the vehicle-side estimation results are calibrated using a complex estimation model deployed in the cloud, achieving accurate SOC estimation. Furthermore, the data uploaded to the cloud is only the data segments related to critical operating conditions, rather than the traditional full data upload. This significantly reduces the amount of data uploaded from the vehicle to the cloud, effectively reducing communication bandwidth pressure. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Schematic diagram of the flow of the battery SOC correction method in an embodiment of the present application; Figure 2 This is a flow chart of step S104 in an embodiment of the present application; Figure 3 Schematic diagram of the flow of a cloud-based battery SOC estimation method in an embodiment of the present application; Figure 4 This is a comparison diagram of the estimation errors between the battery SOC correction method in the embodiment of the present application and the traditional SOC estimation method; Figure 5 Schematic diagram of the structure of the vehicle in the embodiment of the present application. DETAILED DESCRIPTION

[0018] Reference Figure 1 , an embodiment of the present application provides a method for correcting battery SOC, comprising: S101, obtaining real-time operating parameters of a battery; the real-time operating parameters of the battery include: real-time temperature, real-time voltage, and real-time current of the battery; S102, inputting the real-time temperature, real-time voltage, and real-time current of the battery into a pre-built lightweight estimation model to obtain a first SOC estimation value; S103, integrating the real-time current of the battery in ampere-hours to obtain a second SOC estimation value; S104, obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value; S105, when it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, uploading the initial SOC value and a data segment containing the real-time operating parameters of the battery to the cloud for the cloud to determine whether to perform a lightweight estimation network update and issue a corrected SOC value; S106, upon receiving the updated lightweight estimation network and SOC correction value sent from the cloud, updating the lightweight estimation network and obtaining a final SOC value of the battery based on the initial SOC value and the SOC correction value; S107 , when the updated lightweight estimation network and SOC correction value sent by the cloud are not received, the initial SOC value is used as the final SOC value of the battery.

[0019] The battery SOC correction method in the embodiments of the present application is applied to the vehicle side, specifically, the vehicle's battery management system (BMS). The BMS uses, for example, the powerful and resource-optimized STM32H743 MCU as its main control chip. This chip integrates a rich set of peripherals and a high-speed processing core, capable of meeting the complex data processing requirements of the vehicle side.

[0020] The real-time temperature, voltage, and current of the battery in step S101 are all acquired through physical hardware. The real-time voltage, current, and temperature data are acquired as input at a sampling rate of 1 Hz. This data is preprocessed by a signal conditioning circuit to remove noise and outliers, ensuring the accuracy of the input data. For example, an ASE chip is designed into the battery management system. The ASE chip acquires the real-time current of the battery through a current sensor. The current sensor uses a high-precision Hall effect sensor that can accurately measure current changes during the battery's charge and discharge process. The real-time voltage of the battery is acquired through a voltage sensor that uses a voltage divider resistor network combined with a high-precision ADC to achieve high-precision measurement of the battery voltage. The real-time temperature of the battery is acquired through a digital temperature sensor that directly outputs a digital signal, reducing signal conversion errors and ensuring accurate acquisition of battery temperature information.

[0021] In this embodiment of the present application, the lightweight estimation model in step S102 is a LTSM network. This LTSM network accelerates the inference process using an FPGA built into the battery management system (BMS). The FPGA implements parallel computation of the LSTM network through hardware programming, significantly increasing the LSTM network's inference speed and ensuring accurate output of the first SOC estimate within a short period of time.

[0022] In view of the limited computing resources on the vehicle side, knowledge distillation technology is used to migrate the key knowledge in the complex 10-layer deep reinforcement learning model on the cloud to a 2-layer LSTM network. This greatly reduces the complexity and computational complexity of the model while ensuring that the model accuracy loss is less than 0.5%. The complex deep reinforcement learning model on the cloud side serves as the teacher model, and the LSTM network serves as the student model. By adjusting the structure and parameters of the student model, it can imitate the output of the teacher model as much as possible while maintaining a low complexity. After multiple iterative training, the size of the LSTM network is compressed to less than 100KB-150KB, while ensuring that the accuracy loss of the LSTM network relative to the deep reinforcement learning model is less than 0.5%. The inference delay is reduced to less than 10ms-15ms, meeting the vehicle side's requirements for lightweight and real-time models.

[0023] Reference Figure 2 In the embodiment of the present application, step S104 specifically includes: S1041, multiplying the first SOC estimation value by a first preset weight coefficient to obtain a first product value; S1042, multiplying the second SOC estimated value by a second preset weight coefficient to obtain a second product value; S1043, adding the first product value and the second product value to obtain an initial SOC value; The sum of the first preset weight coefficient and the second preset weight coefficient is 1; The first preset weight coefficient and the second preset weight coefficient are determined according to the real-time temperature of the battery.

[0024] The system uses a pre-established lookup table of real-time battery temperature and weight coefficients to quickly and accurately determine the values ​​of the first and second preset weight coefficients k1 and k2 based on the real-time battery temperature. This temperature-adaptive weight adjustment mechanism dynamically optimizes the SOC estimation formula based on the changing characteristics of the battery at different temperatures, further improving estimation accuracy.

[0025] In the embodiment of the present application, the process of establishing the lookup table of the real-time temperature and weight coefficient of the battery includes: First, a representative sample of batteries, spanning different models, batches, and years of use, is selected to ensure a comprehensive representation of actual battery conditions. Charge and discharge experiments are then conducted on N batteries at various ambient temperatures. A series of discrete temperature points, such as -40°C to 150°C, are set at 5°C or 10°C intervals to cover the extreme low and high temperatures the batteries may encounter. During the charge and discharge process at each temperature point, various battery parameters are precisely measured and recorded, including but not limited to precise real-time voltage, current, and SOC measurements, as well as the corresponding real-time temperature.

[0026] Next, the two estimation methods in S102 and S103 are used to estimate the battery SOC of each battery at each temperature point T, and obtain the corresponding first SOC estimation value and second SOC estimation value. ocv and the second SOC estimated value SOC Ah Compared with the actual measured SOC value SOC true Compare and calculate the first SOC error and second SOC error corresponding to each battery at each temperature point T, and then obtain the mean first SOC error of N batteries at each temperature point T. ε OCV ( T ) and the mean of the second SOC error ε Ah ( T ), which satisfies:

[0027] .

[0028] Finally, according to the first SOC mean SOC ocv and the second SOC error mean SOC Ah The first preset weight coefficient k1 and the second preset weight coefficient k2 corresponding to each temperature point T are calculated according to the following formula:

[0029] Repeat the above process to remember the k1 and k2 data corresponding to all test temperature points.

[0030] The temperature points obtained from the experiment and the weight coefficients k1 and k2 determined at the temperature points are organized into a table. One column of the table records the real-time temperature value of the battery, and the other two columns record the corresponding k1 and k2 values ​​respectively.

[0031] To more accurately determine the weight coefficient based on the real-time temperature in practical applications, interpolation may be required within the generated lookup table. For example, when the measured temperature falls between two adjacent temperature points in the lookup table, the corresponding weight coefficient can be estimated using linear interpolation or more complex interpolation methods (such as spline interpolation). Furthermore, changes in the weight coefficient are smoothed to avoid sudden changes due to small temperature changes, thus ensuring the stability of the SOC estimation.

[0032] In step S105, if the operating condition of the battery satisfies at least one of the following conditions: The rate of change of the battery voltage exceeds a first predetermined rate of change; The battery temperature change rate exceeds a second predetermined change rate; The battery switches between charge and discharge states; Determine that the battery's operating conditions fall within predetermined critical conditions.

[0033] Because this predetermined critical operating condition is prone to causing battery SOC errors, the vehicle's computing power is limited and the voltage platform of the lithium iron phosphate battery is flat, making it impossible to self-calibrate the integral error. Therefore, only during highly sensitive moments such as sudden changes in battery voltage, sudden temperature changes, or switching between charge and discharge states, the data segments that are most likely to expose SOC errors are uploaded to the cloud. The cloud-based model with higher complexity and accuracy is used to reduce this SOC error.

[0034] Specifically, in the embodiment of the present application, the first predetermined change rate is, for example, 0.1 V / min, and the second predetermined change rate is, for example, 2° C. / min.

[0035] In an embodiment of the present application, the data segment containing the real-time operating parameters of the battery not only includes the instantaneous data that meets the predetermined critical operating condition triggering moment, but also covers the continuous voltage, current, and temperature sequence from a period of time before the triggering moment (such as the first 30 seconds) to a period of time after the triggering moment (such as the last 10 seconds), so as to comprehensively record the dynamic changes of the battery status under critical operating conditions.

[0036] Furthermore, in an embodiment of the present application, when the battery is in a low SOC or high-rate discharge operating condition, which requires extremely high SOC estimation accuracy, the frequency of uploading data segments is reduced to ensure the accuracy of SOC estimation. Low SOC, for example, refers to an SOC less than 20%, and high rate, for example, refers to an SOC greater than 2C. When the battery is in a low SOC (e.g., less than 20%) or high-rate discharge (e.g., greater than 2C) operating condition, which requires extremely high SOC estimation accuracy, the frequency of uploading data segments is significantly increased from the default 30 minutes / time to 5 minutes / time. In this state, the vehicle side collects and uploads data segments of predetermined key operating conditions more frequently, allowing the cloud to obtain battery status change information more timely, thereby calibrating the SOC more frequently and meeting the demand for high-precision SOC estimation under complex operating conditions.

[0037] In the application embodiment, in order to ensure the security and privacy of data during transmission, differential privacy technology is used for data protection. Before uploading the data segment on the vehicle side, noise is added to the data segment of the predetermined key working condition, and the intensity of the noise is dynamically adjusted according to the sensitivity of the data and the required privacy protection level. At the same time, the MQTT protocol is selected for data transmission. The MQTT protocol has the advantages of being lightweight, low bandwidth consumption, and supporting asynchronous communication. It is particularly suitable for achieving stable data transmission between the vehicle side and the cloud side in scenarios with complex network conditions and limited bandwidth. Among them, the communication module between the vehicle side and the cloud side adopts the NB-IoT module, which supports low-power wide area network (LPWAN) communication and is suitable for scenarios where equipment is widely distributed in the energy storage system. It can achieve stable communication under low bandwidth and low power consumption conditions.

[0038] Furthermore, when the battery is at rest and the SOC is stable (e.g., a 0.01% drop in SOC within 10 minutes), communication between the vehicle and the cloud is suspended to reduce system power consumption and extend device runtime. At this point, the vehicle's battery management system (BMS) retains only basic battery parameter monitoring functions, disables data upload and reception, and enters a low-power sleep state. Upon detecting a change in battery status, such as a battery voltage change exceeding a first predetermined rate, a battery temperature change exceeding a second predetermined rate, or a battery switching between charge and discharge states, the BMS automatically wakes up and resumes normal data collection and communication functions.

[0039] For the cloud, after receiving the data segment of the predetermined key working condition uploaded by the vehicle, it first decrypts and preprocesses the data to remove the added noise and restore the original format of the data. Then, the processed data segment is input into the hybrid model engine, and the third SOC estimation value and the fourth SOC estimation value are obtained through the first estimation model and the second estimation model in the hybrid model engine. The outputs of the two estimation models are merged through a dynamic weighted fusion algorithm to generate the final SOC calibration value. The SOC calibration value is compared with the initial SOC value uploaded by the vehicle, and the error between the two is calculated. If the error between the two exceeds a set ratio (such as 2%) for three consecutive times, it is determined that the current lightweight estimation model on the vehicle needs to be updated.

[0040] When updating the current lightweight estimation model, the cloud leverages multi-node data aggregation and analysis technology to collect data segments uploaded by multiple on-board devices and deeply optimize the LSTM model's weights. Distributed computing and parallel processing accelerate data processing and model training. During the optimization process, algorithms such as stochastic gradient descent are employed to continuously adjust the LSTM model's weights and biases to minimize the error between the LSTM model's first SOC estimate and the actual SOC value. After optimization, the LSTM model is pruned and quantized to remove redundant connections and parameters. The parameters are then represented in a low-precision data format, generating a new lightweight estimation model version that is regularly distributed to the vehicle.

[0041] In the embodiment of the present application, the multi-node data refers to the real-time voltage, real-time current and real-time temperature of each battery cell unit.

[0042] For the cloud, when it determines that the current lightweight estimation model on the vehicle side needs to be updated, it needs to perform weighted processing based on the third SOC estimation value and the fourth SOC estimation value to obtain the SOC calibration value, and then send the SOC calibration value to the vehicle side.

[0043] That is to say, refer to Figure 3 In the embodiment of the present application, the following SOC estimation logic is executed in the cloud: S201, obtaining the initial SOC value and the data segment containing the real-time operating parameters of the battery uploaded by the vehicle; S202, running a first estimation model based on a data segment containing real-time operating parameters of the battery to obtain a third SOC estimation value; S203, running a second estimation model based on the data segment containing the real-time operating parameters of the battery to obtain a fourth SOC estimation value; S204, obtaining an SOC calibration value according to the third SOC estimated value and the fourth SOC estimated value; S205: Determine whether to update the vehicle-side lightweight estimation model based on the SOC calibration value and the SOC initial value; S206, when it is determined that the vehicle-side lightweight estimation model needs to be updated, updating the lightweight estimation model using the SOC calibration value and the data segment containing the real-time operating parameters of the battery; S207: Send the updated lightweight estimation network to the vehicle side, and send the difference between the SOC calibration value and the SOC initial value as the SOC correction value to the vehicle side.

[0044] In step S204, the cloud performs weighted processing based on the third SOC estimated value and the fourth SOC estimated value to obtain the required SOC calibration value. Specifically, the following steps are used: multiplying the third SOC estimated value by a third preset weight coefficient to obtain a third product value; multiplying the fourth SOC estimated value by a fourth preset weight coefficient to obtain a fourth product value; Adding the third product value and the fourth product value to obtain an SOC calibration value; The sum of the third preset weight coefficient and the fourth preset weight coefficient is 1; The third and fourth preset weight coefficients are determined based on the battery's degree of aging and / or current operating conditions to obtain the most accurate SOC calibration value. For example, the first estimation model is an electrochemical model, and the second estimation model is a deep reinforcement learning model. In the early stages of battery aging, the electrochemical model has a higher weight; in the later stages of battery aging, the deep reinforcement learning model gradually increases in weight.

[0045] In the embodiment of the present application, the battery aging degree includes the number of charge and discharge cycles of the battery and the change in the battery internal resistance. The number of charge and discharge cycles of the battery and the change in the battery internal resistance can be regularly uploaded by the vehicle to the cloud.

[0046] The battery's current operating condition includes its current temperature and charge / discharge rate. The current charge / discharge rate is calculated based on the current current uploaded by the vehicle. The charge / discharge rate (C) is defined as the ratio of the charge / discharge current (I) to the battery's rated capacity (Ce), i.e., C = I / Ce. The vehicle's battery management system (BMS) monitors the battery's charge / discharge current in real time, and the battery's rated capacity is known. Given the vehicle's charge / discharge current, the cloud can calculate the current charge / discharge rate through simple calculations.

[0047] In the embodiments of the present application, the first estimation model in the cloud is, for example, an electrochemical model, specifically a simplified version of the P2D model. The electrochemical model is based on the physical and chemical principles within the battery, combined with the actual operating parameters of the battery (such as electrode material properties and electrolyte conductivity). By solving a series of partial differential equations, it makes a preliminary estimate of the battery's SOC, providing an approximate SOC range and boundary constraints based on physical laws.

[0048] In the embodiment of the present application, the first estimation model in the cloud is specifically based on the principle of estimating SOC based on the Nernst equation. The Nernst equation establishes a quantitative relationship between the electrode potential and the concentration of the substances involved in the electrode reaction, and plays a core role in estimating the SOC of secondary ion batteries. The general Nernst equation expression is: ,in: E is the actual measured electrode potential, which can be indirectly determined from the real-time voltage uploaded to the cloud. The electrode potential reflects the actual potential of the electrode in the battery's current state and is a key measurable parameter in the Nernst equation.

[0049] E0 is the standard electrode potential, which depends on the intrinsic properties of the electrode material and is preset in the cloud.

[0050] R is the gas constant, which is 8.314 J·mol -1 ·K -1 .

[0051] T represents the absolute temperature of the battery, which is preset in the cloud.

[0052] n is the number of electron transfers involved in the electrode reaction, and this value is preset in the cloud.

[0053] F is the Faraday constant, which is 96485 C·mol -1 .

[0054] Assuming that the total amount of electrode material is known, the ion content in the current electrode can be determined based on the calculated ratio of the concentration of oxidized substances to the concentration of reduced substances, and the battery's SOC can be further calculated.

[0055] When the Nernst equation is applied to a vehicle's secondary ion battery, the electrode potential E, obtained indirectly through the battery voltage, along with known parameters such as the standard electrode potential E0 and the battery's absolute temperature T, can be used to calculate the ratio of the concentration of oxidized to reduced species. Since the SOC of a secondary ion battery is directly related to the amount of ions embedded / de-intercalated in the electrode, which in turn is closely linked to the ratio of the concentrations of oxidized and reduced species, this method of estimating SOC based on the Nernst equation leverages the fundamental thermodynamic principles of electrochemical reactions within secondary ion batteries. By measuring and calculating the relationship between electrode potential and species concentration, it provides a reliable, physically based approach to SOC estimation. This is used in this electrochemical model to determine the approximate range and boundary constraints of the SOC.

[0056] In the embodiment of the present application, the second estimation model is a deep reinforcement learning model. The deep reinforcement learning model takes historical operating condition data and the currently uploaded data segment as input. Through unsupervised learning of a large amount of historical operating condition data, it mines the complex nonlinear relationships in the battery operation process and performs a more accurate nonlinear fitting of the SOC. In the embodiment of the present application, the actual training and estimation process of the deep reinforcement learning model includes: 1) First, collect a large amount of battery operation history data, which covers various operating parameters of the battery under different working conditions, including but not limited to the time-varying sequence data of the battery's real-time voltage, real-time current, real-time temperature, and the corresponding high-precision SOC measurement value (SOC actualThese data come from a wide range of sources and may include operational records of different types of secondary ion batteries in various application scenarios, such as data collection when electric vehicles are stationary, moving, at high temperatures, or at low temperatures.

[0057] 2) The collected data often has problems such as noise and missing values, so preprocessing is required.

[0058] 3) Use filtering algorithms, such as Gaussian filtering or median filtering, to remove random noise from voltage, current, and temperature data, making the data smoother and reducing the interference of noise on subsequent analysis.

[0059] 4) For data points with missing values, linear interpolation or machine learning-based interpolation methods (such as the K-nearest neighbor algorithm) are used to fill them in according to the time series characteristics of the data to ensure data integrity.

[0060] 5) Map feature data of different ranges (e.g., voltage may be in the range of a few volts, while current may be in the range of milliamperes to amperes) to the same normalized interval, usually [0, 1] or [-1, 1]. Normalization can accelerate model convergence and improve model stability and generalization ability.

[0061] In the embodiments of this application, the deep reinforcement learning model uses a long short-term memory (LSTM) network structure with powerful feature extraction capabilities. LSTM is particularly suitable for processing time series data of battery operation because it can effectively capture long-term dependencies in the data, overcoming the vanishing or exploding gradient problem of traditional methods when processing long time series.

[0062] LSTM networks use gating mechanisms (input gate, forget gate, and output gate) to control the inflow, retention, and output of information. The input gate determines how much information from the current input data is added to the cell state. It is implemented using a sigmoid layer and a tanh layer. The sigmoid layer outputs a value between 0 and 1, representing the importance weight of each dimension of the input data. The tanh layer creates a new vector of candidate values. The two are multiplied together to produce the information that will be added to the cell state. The forget gate determines which information is discarded from the cell state. It is also implemented using a sigmoid layer, with an output value between 0 and 1, where 0 completely discards the information and 1 retains it. The forget gate allows the LSTM to selectively forget information from past cell states, allowing it to better adapt to new input data. The output gate determines which parts of the cell state are output. It consists of a sigmoid layer and a tanh layer. The sigmoid layer determines which parts of the cell state are output, while the tanh layer transforms the cell state and multiplies it with the output of the sigmoid layer to produce the final output.

[0063] When processing battery operating data, the LSTM network sequentially inputs time series data such as the battery's real-time voltage, current, and temperature. Through a gating mechanism, the LSTM network is able to capture the complex patterns and trends of these parameters over time, transforming them into more abstract feature representations. For example, it can learn the unique response patterns of battery voltage and current to changes in SOC at different temperatures. These feature representations contain key information about the battery's operation and lay the foundation for subsequent exploration of complex nonlinear relationships.

[0064] The goal of the LSTM model is to learn an optimal policy to accurately predict SOC so that the predicted value is as close as possible to the actual SOC measurement. This policy is learned by interacting with the environment; at each time step, the model selects an action (predicted SOC value) based on the current state of the environment (i.e., the extracted feature representation). ). The environment will give a reward (r) based on this action. The design of the reward function is crucial, as it guides the model to learn in the right direction. For example, the reward function can be defined as the predicted SOC value Compared with the actual measured value The inverse of the error between: , where ε is a small positive number to prevent the denominator from being zero). Smaller errors result in larger rewards. The model continuously adjusts its parameters to maximize the long-term cumulative reward, gradually learning the complex nonlinear relationship between battery operating parameters and SOC.

[0065] To measure the difference between the predicted value and the actual value, the model uses a loss function. A loss function such as the mean squared error (MSE) loss function is expressed as: Where n0 is the total number of data points in the training dataset, and They are i The predicted SOC value and actual SOC value of each data point. The MSE loss function can intuitively reflect the degree of deviation between the predicted value and the actual value. The greater the deviation, the greater the loss value.

[0066] The model uses a backpropagation algorithm to calculate the gradient of the loss function with respect to the model parameters. Starting with the loss function, the backpropagation algorithm propagates the gradient from the output layer back to each layer of the network using the chain rule, calculating the contribution of each parameter to the loss function. An optimizer (such as stochastic gradient descent (SGD) or Adam) is then used to update the model parameters based on the calculated gradients, continuously reducing the loss function. The Adam optimizer, for example, combines the advantages of the momentum method and the RMSProp algorithm, adaptively adjusting the learning rate of each parameter to more quickly converge to the optimal solution during training. As training progresses, the model gradually adjusts its parameters to better fit the nonlinear relationship between battery operating parameters and SOC, enabling more accurate SOC predictions.

[0067] In summary, the deep reinforcement learning model explores the complex nonlinear relationships in the battery operation process through a series of steps such as battery operation data collection, preprocessing, feature extraction, strategy learning, and model optimization, achieving high-precision nonlinear fitting of the battery SOC, and providing key support for the end-cloud collaborative battery SOC dynamic calibration system.

[0068] In an embodiment of the present application, for example, an AWS EC2 P4 instance or an Alibaba Cloud ECS g6 instance is used as a cloud server. The instance is equipped with a powerful GPU that can accelerate the training and calculation of deep learning models.

[0069] On cloud servers, deploy model services based on the PyTorch deep learning framework or the TensorFlow deep learning framework, and leverage PyTorch's automatic differentiation and distributed training features or TensorFlow's powerful model training and optimization capabilities to achieve efficient training and optimization of hybrid models.

[0070] A data storage and management system was established on cloud servers to store and manage vehicle-side lightweight estimation model parameters and data segments uploaded from multiple vehicles, ensuring secure data storage and rapid retrieval. The data management system utilizes Alibaba Cloud's database service, supporting large-scale data storage and rapid retrieval, and efficiently managing key operating condition data segments and model parameters from multiple energy storage units.

[0071] Reference Figure 4 , applying the above method in the embodiment of the present application, the batteries of multiple electric vehicles were tested under low temperature conditions of -20°C; the results showed that the SOC estimation error was significantly reduced from 8.2% of the traditional method to 2.1%, which greatly improved the accuracy of SOC estimation in low temperature environments and provided more reliable data support for energy management and cruising range prediction of electric vehicles under low temperature conditions.

[0072] By adopting a mechanism for dynamically uploading data segments to the cloud, data upload volume is reduced by 76% compared to the traditional method of uploading all data to the cloud. By only uploading data segments relevant to predetermined critical operating conditions, communication bandwidth pressure is effectively reduced. In actual testing, communication overhead was less than 50KB per hour, achieving efficient data transmission under low-bandwidth conditions. Furthermore, the optimized configuration of the MQTT protocol ensures the stability and reliability of data transmission, achieving a data transmission success rate exceeding 99% in complex network environments.

[0073] This embodiment achieves dynamic SOC calibration through collaboration between the vehicle and the cloud. This allows LFP batteries to maintain SOC estimation errors within 2.5% under various temperature and charge / discharge rate conditions, significantly outperforming traditional estimation methods (typically with errors between 5% and 10%). By employing a dynamic data segment upload mechanism, data upload volume is reduced by 80% compared to traditional full data upload methods. In actual testing, communication overhead was less than 30KB per hour, enabling efficient data transmission under low-bandwidth conditions. Optimized configuration of the NB-IoT module and the MQTT protocol ensures stable and reliable data transmission. Even in complex network environments, the data transmission success rate exceeds 98%. In the energy storage system, when the battery is in predetermined critical operating conditions, such as low SOC (less than 15%) or high-rate charge / discharge (greater than 2C), it automatically switches to high-frequency calibration mode (increasing data upload frequency to every 10 minutes), ensuring high-accuracy and real-time SOC estimation. When the battery is at rest and the SOC is stable, it automatically enters sleep mode, reducing system power consumption and extending device operation time.

[0074] The present application also provides a battery SOC correction system, comprising: a controller and a vehicle-mounted network terminal connected to each other; the controller is configured to: Acquire real-time operating parameters of the battery; the real-time operating parameters of the battery include: real-time temperature, real-time voltage and real-time current of the battery; Inputting the real-time temperature, real-time voltage, and real-time current of the battery into a pre-built lightweight estimation model to obtain a first SOC estimation value; Integrate the real-time current of the battery in ampere hours to obtain a second SOC estimation value; Obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value; When it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, the initial SOC value and a data segment containing the real-time operating parameters of the battery are uploaded to the cloud via the vehicle-mounted network terminal, so that the cloud can determine whether to perform a lightweight estimation network update and send a corrected SOC value; When the updated lightweight estimation network and SOC correction value sent from the cloud are received through the vehicle-mounted network terminal, the lightweight estimation network is updated, and the final SOC value of the battery is obtained based on the initial SOC value and the SOC correction value; When the updated lightweight estimation network and SOC correction value sent from the cloud are not received through the vehicle-mounted network terminal, the initial SOC value is used as the final SOC value of the battery.

[0075] In the embodiments of the present application, the controller is, for example, a vehicle-mounted battery management system (BMS) or vehicle-mounted controller (VCU). Depending on actual product requirements, either an all-in-one domain controller or a separate domain controller on the vehicle can be used as the aforementioned controller. The controller can autonomously collect battery current, temperature, and voltage using the ASE chip described in the aforementioned embodiments (the collection principles of the ASE chip are well known in the art).

[0076] In addition, the in-vehicle connected terminal refers to the T-Box (Telematics Box) on the vehicle that is responsible for information interaction and communication with the cloud.

[0077] This application also provides a cloud-based battery SOC estimation device, including: The acquisition module is used to obtain the initial SOC value uploaded by the vehicle and the data segment containing the real-time operating parameters of the battery; a first estimation module, configured to execute a first estimation model based on a data segment containing real-time operating parameters of the battery to obtain a third SOC estimation value; a second estimation module, configured to run a second estimation model based on a data segment containing real-time operating parameters of the battery to obtain a fourth SOC estimation value; an SOC calibration value estimating module, configured to obtain an SOC calibration value according to the third SOC estimation value and the fourth SOC estimation value; A model update judgment module, configured to determine whether to update the vehicle-side lightweight estimation model based on the SOC calibration value and the SOC initial value; a model updating module, configured to update the lightweight estimation model on the vehicle side using the SOC calibration value and a data segment containing real-time operating parameters of the battery when determining that the lightweight estimation model on the vehicle side needs to be updated; The sending module is used to send the updated lightweight estimation network to the vehicle side, and send the difference between the SOC calibration value and the SOC initial value as the SOC correction value to the vehicle side.

[0078] The specific principle of the above-mentioned SOC cloud estimation device can refer to the working principle of the above-mentioned SOC cloud estimation method.

[0079] Figure 5 FIG2 is a block diagram illustrating a vehicle 200 according to an exemplary embodiment. For example, vehicle 200 may be a hybrid vehicle, a non-hybrid vehicle, an electric vehicle, a fuel cell vehicle, or another type of vehicle. Vehicle 200 may be an autonomous vehicle, a semi-autonomous vehicle, or a non-autonomous vehicle.

[0080] Reference Figure 5 Vehicle 200 may include various subsystems, such as an infotainment system 210, a perception system 220, a decision control system 230, a drive system 240, and a computing platform 250. Vehicle 200 may also include more or fewer subsystems, and each subsystem may include multiple components. Furthermore, each subsystem and each component of vehicle 200 may be interconnected via wired or wireless means. In some embodiments, infotainment system 210 may include a communication system, an entertainment system, and a navigation system.

[0081] The perception system 220 may include several sensors for sensing information about the environment surrounding the vehicle 200. For example, the perception system 220 may include a global positioning system (which may be a GPS system, a BeiDou system, or other positioning systems), an inertial measurement unit (IMU), a laser radar, a millimeter-wave radar, an ultrasonic radar, and a camera.

[0082] The decision control system 230 may include a computing system, a vehicle controller, a steering system, a throttle, and a braking system. The drive system 240 may include components that provide power and movement for the vehicle 200. In one embodiment, the drive system 240 may include an engine, an energy source, a transmission system, and wheels. The engine may be an internal combustion engine, an electric motor, or an air compression engine, or a combination thereof. The engine is capable of converting energy provided by the energy source into mechanical energy.

[0083] Some or all functions of the vehicle 200 are controlled by a computing platform 250. The computing platform 250 may include at least one processor 251 and a memory 252. The processor 251 may execute instructions 253 stored in the memory 252.

[0084] The processor 251 can be any conventional processor, such as a commercially available CPU. The processor can also include a graphics processor (GPU), a field programmable gate array (FPGA), a system on chip (SOC), an application specific integrated circuit (ASIC), or a combination thereof.

[0085] The memory 252 may be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0086] In addition to instructions 253 , memory 252 may also store data, such as road maps, route information, and data on the vehicle's location, direction, speed, etc. The data stored in memory 252 may be used by computing platform 250 .

[0087] In the embodiment of the present disclosure, the processor 251 may execute the instruction 253 to complete all or part of the steps of the above-mentioned vehicle control method.

[0088] The present disclosure also provides a computer-readable storage medium having computer program instructions stored thereon. When the program instructions are executed by a processor, the steps of the vehicle control method provided by the present disclosure are implemented.

[0089] Furthermore, the word "exemplary" is used herein to mean serving as an example, instance, or illustration. Any aspect or design described herein as "exemplary" is not necessarily to be construed as advantageous over other aspects or designs. Rather, the use of the word exemplary is intended to present concepts in a concrete manner. As used herein, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified or clear from the context, "X applies to A or B" is intended to mean any of the natural inclusive permutations. That is, if X applies to A; X applies to B; or X applies to both A and B, then "X applies to A or B" satisfies any of the aforementioned instances. Furthermore, the articles "a" and "an," as used in this application and the appended claims, are generally understood to mean "one or more," unless otherwise specified or clear from the context to refer to the singular form.

[0090] Likewise, although the present disclosure has been shown and described with respect to one or more implementations, equivalent variations and modifications will occur to those skilled in the art after reading and understanding the specification and drawings. The present disclosure includes all such modifications and variations and is limited only by the scope of the claims. In particular, with respect to the various functions performed by the components described above (e.g., elements, resources, etc.), unless otherwise indicated, the terms used to describe such components are intended to correspond to any component (functionally equivalent) that performs the specific functions of the described components, even if structurally not equivalent to the disclosed structures. In addition, although specific features of the present disclosure may have been disclosed with respect to only one of several implementations, such features may be combined with one or more other features of other implementations as may be desired and beneficial for any given or specific application. In addition, with respect to the terms "including," "having," "having," "having," or variations thereof used in the specific embodiments or claims, such terms are intended to be inclusive in a manner similar to the term "comprising."

[0091] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the following claims.

[0092] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

[0093] It should be noted that the terms "first," "second," and the like in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the numbers used in this manner are interchangeable where appropriate so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure as detailed in the appended claims.

[0094] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with an embodiment or example is included in at least one embodiment or example of the present disclosure. In this specification, the illustrative use of the above terms does not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0095] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present disclosure belong.

[0096] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processing module, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection having one or more wires (control method), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or otherwise processing it in a suitable manner if necessary, and then storing it in a computer memory.

[0097] It should be understood that the various parts of the embodiments of the present disclosure can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0098] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0099] In addition, the functional units in the various embodiments of the present disclosure may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into a module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium. The above-mentioned storage medium may be a read-only memory, a magnetic disk, an optical disk, etc.

[0100] Although the embodiments of the present disclosure have been shown and described above, it is understood that the above embodiments are exemplary and are not to be construed as limitations on the present disclosure. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present disclosure.

Claims

1. A cloud-based battery SOC estimation method, characterized in that: include: Obtain the initial SOC value and the data segment containing the real-time operating parameters of the battery uploaded by the vehicle; Running a first estimation model based on a data segment containing real-time operating parameters of the battery to obtain a third SOC estimation value; Running a second estimation model based on a data segment containing real-time operating parameters of the battery to obtain a fourth SOC estimation value; Obtaining an SOC calibration value according to the third SOC estimated value and the fourth SOC estimated value; Determining whether to update a vehicle-side lightweight estimation model according to the SOC calibration value and the SOC initial value; When it is determined that the vehicle-side lightweight estimation model needs to be updated, the lightweight estimation model is updated using the SOC calibration value and the data segment containing the real-time operating parameters of the battery; The updated lightweight estimation network is sent to the vehicle side, and the difference between the SOC calibration value and the SOC initial value is sent to the vehicle side as the SOC correction value.

2. The cloud-based battery SOC estimation method according to claim 1, characterized in that: The step of obtaining an SOC calibration value according to the third SOC estimated value and the fourth SOC estimated value includes: multiplying the third SOC estimated value by a third preset weight coefficient to obtain a third product value; multiplying the fourth SOC estimated value by a fourth preset weight coefficient to obtain a fourth product value; Adding the third product value and the fourth product value to obtain an SOC calibration value; The sum of the third preset weight coefficient and the fourth preset weight coefficient is 1; The third preset weight coefficient and the fourth preset weight coefficient are determined according to the aging degree of the battery and / or the current working condition of the battery.

3. A method for correcting battery SOC, characterized in that: include: Obtain real-time operating parameters of the battery; The real-time operating parameters of the battery include: real-time temperature, real-time voltage and real-time current of the battery; Inputting the real-time temperature, real-time voltage, and real-time current of the battery into a pre-built lightweight estimation model to obtain a first SOC estimation value; Integrate the real-time current of the battery in ampere hours to obtain a second SOC estimation value; Obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value; When it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, the initial SOC value and a data segment containing the real-time operating parameters of the battery are uploaded to the cloud for the cloud to determine whether to perform a lightweight estimation network update and send a corrected SOC value; Upon receiving the updated lightweight estimation network and SOC correction value sent from the cloud, updating the lightweight estimation network and obtaining the final SOC value of the battery based on the initial SOC value and the SOC correction value; When the updated lightweight estimation network and SOC correction value sent by the cloud are not received, the initial SOC value is used as the final SOC value of the battery; The cloud uses the cloud-based battery SOC estimation method according to claim 1 or 2 to determine the updated lightweight estimation network and SOC correction value.

4. The battery SOC correction method according to claim 3, characterized in that: If the battery's operating conditions meet at least one of the following conditions: The rate of change of the battery voltage exceeds a first predetermined rate of change; The battery temperature change rate exceeds a second predetermined change rate; The battery switches between charge and discharge states; Determine that the battery's operating conditions fall within predetermined critical conditions.

5. The battery SOC correction method according to claim 3, characterized in that: The step of obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value includes: multiplying the first SOC estimation value by a first preset weight coefficient to obtain a first product value; multiplying the second SOC estimated value by a second preset weight coefficient to obtain a second product value; Adding the first product value and the second product value to obtain an initial SOC value; The sum of the first preset weight coefficient and the second preset weight coefficient is 1; The first preset weight coefficient and the second preset weight coefficient are determined according to the real-time temperature of the battery.

6. The battery SOC correction method according to claim 3, characterized in that: The step of obtaining a final SOC value of the battery according to the initial SOC value and the corrected SOC value includes: The initial SOC value and the corrected SOC value are added to obtain a final SOC value of the battery.

7. The battery SOC correction method according to claim 3, characterized in that: When it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, uploading the initial SOC value and the data segment containing the real-time operating parameters of the battery to the cloud includes: When the SOC of the battery is greater than or equal to a preset SOC and the discharge rate of the battery is less than or equal to a preset discharge rate, uploading the data segment to the cloud at a first preset frequency; When the SOC of the battery is less than a preset SOC or the discharge rate of the battery is greater than a preset discharge rate, uploading the data segment to the cloud at a second preset frequency; The first preset frequency is lower than the second preset frequency.

8. A battery SOC correction system, characterized in that: include: Connected controllers and vehicle-mounted network terminals; The controller is used to: Obtain real-time operating parameters of the battery; The real-time operating parameters of the battery include: real-time temperature, real-time voltage and real-time current of the battery; Inputting the real-time temperature, real-time voltage, and real-time current of the battery into a pre-built lightweight estimation model to obtain a first SOC estimation value; Integrate the real-time current of the battery in ampere hours to obtain a second SOC estimation value; Obtaining an initial SOC value according to the first SOC estimated value and the second SOC estimated value; When it is identified that the operating condition of the battery belongs to a predetermined critical operating condition, the initial SOC value and a data segment containing the real-time operating parameters of the battery are uploaded to the cloud via the vehicle-mounted network terminal, so that the cloud can determine whether to perform a lightweight estimation network update and send a corrected SOC value; When the updated lightweight estimation network and SOC correction value sent from the cloud are received through the vehicle-mounted network terminal, the lightweight estimation network is updated, and the final SOC value of the battery is obtained based on the initial SOC value and the SOC correction value; When the updated lightweight estimation network and SOC correction value sent from the cloud are not received through the vehicle-mounted network terminal, the initial SOC value is used as the final SOC value of the battery; The cloud uses the cloud-based battery SOC estimation method according to claim 1 or 2 to determine the updated lightweight estimation network and SOC correction value.

9. A cloud-based battery SOC estimation device, characterized in that: include: The acquisition module is used to obtain the initial SOC value uploaded by the vehicle and the data segment containing the real-time operating parameters of the battery; a first estimation module, configured to execute a first estimation model based on a data segment containing real-time operating parameters of the battery to obtain a third SOC estimation value; a second estimation module, configured to run a second estimation model based on a data segment containing real-time operating parameters of the battery to obtain a fourth SOC estimation value; an SOC calibration value estimating module, configured to obtain an SOC calibration value according to the third SOC estimation value and the fourth SOC estimation value; A model update judgment module, configured to determine whether to update the vehicle-side lightweight estimation model based on the SOC calibration value and the SOC initial value; a model updating module, configured to update the lightweight estimation model on the vehicle side using the SOC calibration value and a data segment containing real-time operating parameters of the battery when determining that the lightweight estimation model on the vehicle side needs to be updated; The sending module is used to send the updated lightweight estimation network to the vehicle side, and send the difference between the SOC calibration value and the SOC initial value as the SOC correction value to the vehicle side.

10. A vehicle, characterized in that: A battery SOC correction system comprising the battery SOC correction system according to claim 8.

Images