Method and device for predicting residual cycle life of battery, medium and equipment

Abstract

The invention belongs to the technical field of battery management, and provides a battery residual cycle life prediction method and device, a medium and equipment, and the method comprises the steps: carrying out the prediction of a battery state at a next moment through a neural network based on the battery state data at a plurality of historical moments, and carrying out the calculation of the residual cycle life of a battery through an SPM model according to the predicted SOC and SOH; calculating to obtain voltage, current and internal resistance in the battery; if the voltage, the current and the internal resistance calculated by the SPM model are consistent with those predicted by the neural network, the SPM model adjusts the physical and chemical parameters of the battery according to the battery state at the next moment predicted by the neural network; otherwise, after fine tuning training is carried out on the neural network, the battery state at the next moment is predicted again; and calculating the remaining cycle life of the battery based on the battery state at the next moment predicted by the neural network. And the accuracy of residual cycle life prediction is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery management, and in particular relates to a method, device, medium and equipment for predicting the remaining cycle life of a battery. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] The remaining cycle life of a power battery refers to how many charge and discharge cycles the battery can go through without failing under the current battery health condition.

[0004] Accurately predicting the remaining useful life (RUL) of a battery is crucial for battery management, maintenance, and replacement. Existing methods usually require a detailed understanding of the internal structure and material properties of the battery, which is time-consuming and susceptible to uncertain factors such as temperature, operating conditions, and SOC. However, most current methods use neural networks to predict battery SOC and other data. This method is more affected by the previous training data, and the error in the prediction process is uncontrollable, which can easily lead to a decrease in the accuracy of the remaining cycle life prediction. Summary of the invention

[0005] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a battery remaining cycle life prediction method, device, medium and equipment. Through the interaction of a neural network and a semi-physical model (Semi-Physical Model, SPM), not only the monitoring of the battery state prediction error is realized, but also the neural network and SPM model can be continuously optimized to ensure the prediction accuracy of the model and improve the accuracy of the remaining cycle life prediction.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] A first aspect of the present invention provides a method for predicting the remaining cycle life of a battery.

[0008] A method for predicting remaining cycle life of a battery, comprising:

[0009] Obtaining battery status at several historical moments, wherein the battery status includes voltage, current, SOC, SOH, and internal resistance;

[0010] Based on the battery status data of several historical moments, after the battery status at the next moment is predicted through the neural network, the SPM model calculates the voltage, current and internal resistance of the battery according to the predicted SOC and SOH; if the voltage, current and internal resistance calculated by the SPM model are consistent with those predicted by the neural network, the SPM model adjusts the battery physical and chemical parameters according to the battery status at the next moment predicted by the neural network; otherwise, the voltage, current and internal resistance calculated by the SPM model and the SOC and SOH predicted by the LSTM neural network are used as a set of training data, and the neural network is fine-tuned and trained to re-predict the battery status at the next moment;

[0011] The remaining cycle life of the battery is calculated based on the battery status at the next moment predicted by the neural network.

[0012] Furthermore, the SPM model calculates the voltage inside the battery according to the predicted SOC, and the specific steps include:

[0013] Based on the predicted SOC, the voltage is simulated through the corresponding relationship curve between voltage and SOC;

[0014] If the error between the simulated voltage and the voltage at the next moment predicted by the neural network exceeds the set value, the simulated voltage is used as the voltage output by the SPM model; otherwise, the voltage at the next moment predicted by the neural network is used as the voltage output by the SPM model.

[0015] Furthermore, the SPM model calculates the internal resistance of the battery according to the predicted SOH, and the specific steps include:

[0016] Calculate the ratio of the next moment's internal resistance predicted by the neural network to the reference internal resistance, and simulate the battery's SOH;

[0017] If the error between the simulated SOH and the SOH at the next moment predicted by the neural network exceeds the set value, the internal resistance of the battery is calculated based on the SOH at the next moment predicted by the neural network and the reference internal resistance as the internal resistance output by the SPM model; otherwise, the internal resistance at the next moment predicted by the neural network is used as the internal resistance output by the SPM model.

[0018] Furthermore, the voltage and internal resistance at the historical moment are obtained by extracting features from the high-frequency pulse current.

[0019] A second aspect of the present invention provides a battery remaining cycle life prediction device.

[0020] A battery remaining cycle life prediction device, comprising:

[0021] A data acquisition module is configured to: acquire battery status at several historical moments, wherein the battery status includes voltage, current, SOC, SOH and internal resistance;

[0022] The interactive prediction module is configured as follows: based on the battery status data of several historical moments, after predicting the battery status at the next moment through the neural network, the SPM model calculates the voltage, current and internal resistance inside the battery according to the predicted SOC and SOH; if the voltage, current and internal resistance calculated by the SPM model are consistent with those predicted by the neural network, the SPM model adjusts the battery physical and chemical parameters according to the battery status at the next moment predicted by the neural network; otherwise, based on the voltage, current and internal resistance calculated by the SPM model and the SOC and SOH predicted by the LSTM neural network as a set of training data, the neural network is fine-tuned and trained, and the battery status at the next moment is re-predicted;

[0023] The life calculation module is configured to calculate the remaining cycle life of the battery based on the battery state at the next moment predicted by the neural network.

[0024] Furthermore, the SPM model calculates the voltage inside the battery according to the predicted SOC, and the specific steps include:

[0025] Based on the predicted SOC, the voltage is simulated through the corresponding relationship curve between voltage and SOC;

[0026] If the error between the simulated voltage and the voltage at the next moment predicted by the neural network exceeds the set value, the simulated voltage is used as the voltage output by the SPM model; otherwise, the voltage at the next moment predicted by the neural network is used as the voltage output by the SPM model.

[0027] Furthermore, the SPM model calculates the internal resistance of the battery according to the predicted SOH, and the specific steps include:

[0028] Calculate the ratio of the next moment's internal resistance predicted by the neural network to the reference internal resistance, and simulate the battery's SOH;

[0029] If the error between the simulated SOH and the SOH at the next moment predicted by the neural network exceeds the set value, the internal resistance of the battery is calculated based on the SOH at the next moment predicted by the neural network and the reference internal resistance as the internal resistance output by the SPM model; otherwise, the internal resistance at the next moment predicted by the neural network is used as the internal resistance output by the SPM model.

[0030] Furthermore, the voltage and internal resistance at the historical moment are obtained by extracting features from the high-frequency pulse current.

[0031] A third aspect of the present invention provides a computer-readable storage medium.

[0032] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in a method for predicting remaining cycle life of a battery as described in the first aspect above.

[0033] A fourth aspect of the present invention provides a computer device.

[0034] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in a method for predicting the remaining cycle life of a battery as described in the first aspect above are implemented.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention not only realizes the monitoring of battery state prediction error through the interaction of the neural network and the SPM model, but also can continuously optimize the neural network and the SPM model, ensure the prediction accuracy of the model, and improve the accuracy of the remaining cycle life prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 is a flow chart of a method for predicting remaining cycle life of a battery shown in Embodiment 1 of the present invention;

[0039] Figure 2 is a schematic diagram of the interaction between the neural network and the SPM model shown in the first embodiment of the present invention;

[0040] Figure 3 It is a structural diagram of a computer device shown in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0042] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0043] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0044] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the methods and devices according to various embodiments of the present invention. It should be noted that each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the module, the program segment, or a part of the code may include one or more executable instructions for implementing the logical functions specified in the various embodiments. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the flowchart and / or block diagram, and the combination of boxes in the flowchart and / or block diagram can be implemented using a dedicated hardware-based device that performs a specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0045] Embodiment 1

[0046] This embodiment provides a method for predicting the remaining cycle life of a battery.

[0047] The accuracy and efficiency of prediction can be improved by combining the RUL estimation method of Long Short-Term Memory (LSTM) and Semi-Physical Model (SPM). This method combines the advantages of data-driven and physical models, and improves the accuracy and robustness of prediction by combining specific learning algorithms and physical models.

[0048] This embodiment provides a method for predicting the remaining cycle life of a battery. First, the power battery data on the vehicle side is collected and updated in real time to a big database, such as voltage, current, SOC, SOH, high-frequency pulse current (HPPC) and other data; next, the big database directly imports the multi-dimensional data into the cloud, and the cloud processes and predicts the data based on LSTM and SPM models.

[0049] A method for predicting the remaining cycle life of a battery provided in this embodiment includes: obtaining battery status data (including voltage, current, SOC, SOH, and R) at several historical moments; predicting the battery status (voltage, current, SOC, SOH, and R) at the next moment through an LSTM neural network based on the battery status data at the several historical moments, and then using an SPM model to calculate the voltage, current, and internal resistance inside the battery according to the predicted SOC and SOH; if the voltage, current, and internal resistance calculated by the SPM model are consistent with those predicted by the LSTM neural network, the SPM model adjusts the physical and chemical parameters of the battery according to the battery status at the next moment predicted by the neural network; otherwise, using the voltage, current, and internal resistance calculated by the SPM model and the SOC and SOH predicted by the LSTM neural network as a set of training data, the LSTM neural network is fine-tuned and trained, and then the battery status at the next moment is re-predicted; and the battery status at the next moment predicted by the LSTM neural network is calculated to calculate the remaining cycle life of the battery.

[0050] Among them, the SPM model calculates the voltage inside the battery according to the predicted SOC. The specific steps include: based on the predicted SOC, simulating the voltage through the corresponding relationship curve between voltage and SOC; if the error between the simulated voltage and the voltage at the next moment predicted by the LSTM neural network exceeds the set value, the simulated voltage is used as the voltage output by the SPM model, otherwise, the voltage at the next moment predicted by the LSTM neural network is used as the voltage output by the SPM model.

[0051] Among them, the SPM model calculates the internal resistance of the battery according to the predicted SOH. The specific steps include: calculating the ratio of the internal resistance at the next moment predicted by the neural network to the reference internal resistance, and simulating the SOH of the battery; if the error between the simulated SOH and the SOH at the next moment predicted by the LSTM neural network exceeds the set value, then based on the SOH at the next moment predicted by the LSTM neural network and the reference internal resistance, calculate the internal resistance of the battery as the internal resistance output by the SPM model; otherwise, use the internal resistance at the next moment predicted by the LSTM neural network as the internal resistance output by the SPM model.

[0052] A battery remaining cycle life prediction method provided in this embodiment is applied to a battery of a certain PHEV model: voltage range: 2.5V-4.2V, current range: -100A-100A, temperature range: -20℃-60℃, SOC range: 0%-100%, SOH range: 60%-100%.

[0053] This embodiment provides a method for predicting the remaining cycle life of a battery. Figure 1 As shown, the following steps are included:

[0054] Step 1: Data collection and preprocessing.

[0055] Step 101, real-time data collection: collect battery data in real time through the battery management software in the vehicle-side controller, including voltage V, current I, SOC, SOH, high-frequency pulse current (HPPC), etc.

[0056] Step 102, HPPC data processing: extract features from the high-frequency pulse current (HPPC) test data, including the internal resistance R of the battery cell, voltage response, etc. The specific steps are as follows:

[0057] Preprocess the HPPC data to filter out noise and outliers;

[0058] Calculate the change in internal resistance of the battery cell under high-frequency pulses and extract the voltage response characteristics;

[0059] The processed HPPC features are used as one of the input data of the model.

[0060] Step 103, anomaly detection and preprocessing: Apply anomaly detection algorithms to filter abnormal data in real time. Perform Z-value centering and normalization on the data to speed up model training and optimization.

[0061] Step 2: Data transmission and storage.

[0062] Step 201, edge computing node: deploy edge computing nodes on the vehicle side to perform preliminary data processing to reduce the transmission burden. The processed data is transmitted to the vehicle T-box via the CAN network.

[0063] Step 202, cloud storage and processing: T-box transfers data to a large database, extracts features from the data in the database, and directly imports the data into the cloud model for processing.

[0064] Step 3: LSTM model and SPM training and optimization.

[0065] Step 301, construct a hybrid model architecture: Figure 2 As shown, LSTM is combined with the battery physical model (SPM), LSTM is used to process time series data, and the SPM model is used to simulate the physical and chemical behavior of the battery.

[0066] LSTM model: Based on the time series input data of voltage, current, SOC, SOH, R, etc., predict the battery status (voltage, current, SOC, SOH, R) at the next moment. The input of LSTM is [samples, timesteps, features], where samples represents the amount of data, timesteps represents the time step, and features represents the feature dimension.

[0067] LSTM example parameters: input data length: 100 time steps; feature dimension: 5 (voltage, current, SOC, SOH, internal resistance); number of LSTM layers: 3 layers, 128 units per layer.

[0068] SPM model: The SPM model (Semi-Physical Model) combines the physical properties of the battery and the data-driven method. It can simulate the voltage response and capacity change of the battery and provide a mechanism analysis of battery aging.

[0069] SPM parameters: internal resistance: dynamic range from 10mΩ to 200mΩ; voltage-capacity curve: fitted by experimental data, based on different SOC points; electrode reaction kinetics: based on the Butler-Volmer equation.

[0070] Step 302: Coupling and interaction between LSTM and SPM models:

[0071] (1) Use the training set to train the LSTM and SPM models respectively:

[0072] Initialize the LSTM model and SPM model parameters. The input of the LSTM model is the historical operation data of the battery (voltage, current, SOC, SOH, R, etc.), and the input of the SPM model is the physical and chemical characteristic parameters of the battery.

[0073] Use the training set data to train the LSTM model, extract time series features, and update the model weights. The LSTM model can learn the behavioral characteristics of the battery under different working conditions and provide time series prediction results.

[0074] LSTM model training details: data is processed in batches with a batch size of 32; each training iteration is 100 rounds with a learning rate of 0.001; cross-validation is used to ensure the generalization ability of the model.

[0075] The SPM model is used to simulate the same data to obtain the physical and chemical behavior characteristics of the battery. The SPM model can simulate the physical and chemical processes inside the battery in detail and provide the response characteristics of the battery under different aging conditions.

[0076] SPM model simulation details: the initial SOC is set to 50%, and the voltage changes under different charge and discharge rates are simulated; the internal resistance change range is set to 20mΩ to 150mΩ; the time step of each simulation is 1 second, and the total simulation time is 3600 seconds.

[0077] (2) The prediction results of the LSTM model are used as the input of the SPM model for secondary correction. Specifically, the battery state change results predicted by the LSTM model are input into the SPM model, and the SPM model further simulates the physical and chemical behavior of the battery to correct the error of the LSTM prediction. This step is specifically divided into the following sub-steps:

[0078] (101) The SOC and SOH results of the next moment predicted by the LSTM model are taken as input and passed to the SPM model;

[0079] (102) The SPM model calculates the voltage change, current response, internal resistance, and corresponding capacity change inside the battery based on the SOC and SOH values ​​input by the LSTM;

[0080] (103) The SPM model feeds back the calculated physicochemical properties such as voltage and current to the LSTM model to correct the prediction results of the LSTM model. Correction process details: According to the voltage response calculated by SPM, the SOC value predicted by LSTM is adjusted, and the calculation error range is ±2%; combined with the internal resistance change simulated by SPM, the SOH value predicted by LSTM is corrected, and the calculation error range is ±1%.

[0081] The steps of adjusting the SOC value according to the voltage response are as follows:

[0082] a) Voltage response simulation: First, under the set initial conditions (such as SOC = 50%, different charge and discharge rates and different internal resistance changes), the SPM model was used to simulate the voltage response of the battery with a time step of 1 second and a total simulation time of 3600 seconds.

[0083] b) Establishment of voltage-SOC relationship: Through simulation, the change of voltage response over time can be obtained. Since the open circuit voltage (OCV) of the battery is directly related to the SOC, a corresponding relationship curve between voltage and SOC can be established. As time goes by, the change in voltage reflects the change in lithium ion concentration inside the battery, and the change in lithium ion concentration directly affects the SOC value.

[0084] c) Error calculation and SOC adjustment: The simulated voltage response is compared with the actual measured voltage (here refers to the voltage at the next moment predicted by the LSTM neural network). If the simulated voltage deviates from the actual measured voltage, this means that the current SOC estimate may be inaccurate. By minimizing the voltage error, the SOC value can be adjusted in the opposite direction. Specifically, if the simulated voltage is higher than the actual voltage, it may indicate that the SOC is overestimated and needs to be reduced; conversely, if the simulated voltage is lower than the actual voltage, it means that the SOC is underestimated and needs to be increased.

[0085] d) Dynamic correction: The SOC is adjusted continuously in an iterative manner until the simulated voltage matches the actual measured voltage. This method combines the physical and chemical properties of the battery (simulated by the SPM model) and the actual voltage response to dynamically correct the SOC.

[0086] The SPM model mainly simulates the behavior of single particles inside the battery, describing the diffusion and reaction process of lithium ions in the electrode particles. SPM assumes that each electrode is composed of countless uniform small particles, so the physical and chemical behavior of the electrode can be predicted through simplified mass and charge balance equations.

[0087] In the SPM model, the voltage response is obtained by calculating the polarization voltage, ohmic voltage drop, and the effect of solid phase diffusion of the battery. The specific calculation process is as follows:

[0088] Lithium ion diffusion in particles: SPM predicts the electrochemical reaction rate of the electrode by solving the change in lithium ion concentration within the particles. The change in lithium ion concentration will affect the voltage response of the battery.

[0089] Polarization voltage: The electrochemical reaction rate is related to the electrode surface concentration. The polarization voltage can be calculated using the Butler-Volmer equation, which reflects the reaction kinetics inside the battery.

[0090] Ohmic voltage drop: Ohmic voltage drop is the resistance generated when the battery current passes through the electrodes and electrolyte. The internal resistance set in the simulation (20mΩ to 150mΩ) will affect the ohmic voltage drop.

[0091] Total voltage: By adding the ohmic voltage drop and polarization voltage to the open circuit voltage (OCV), you can get the voltage response of the battery. The open circuit voltage is usually a function of SOC, so the OCV at different SOCs will be different.

[0092] The steps for adjusting SOH according to the change in internal resistance calculated by SPM are as follows:

[0093] a) Simulation and monitoring of internal resistance:

[0094] The SPM model is used to simulate the battery, set different aging states and cycle numbers, simulate the voltage response of the battery at different aging stages, and calculate the internal resistance of the battery.

[0095] The change in internal resistance can be calculated by comparing the voltage drop and charge and discharge current under different charge and discharge cycles. Generally, an increase in internal resistance will cause the battery's voltage response curve to become steeper and the discharge voltage to drop faster.

[0096] b) Relationship between internal resistance change and SOH:

[0097] Baseline internal resistance: For a new battery, the baseline value of internal resistance is known and is usually measured in its initial state. In a healthy battery, the internal resistance is low.

[0098] SOH estimation formula: SOH is usually characterized by the relative increase in internal resistance. A commonly used SOH estimation formula is:

[0099]

[0100] c) Dynamic adjustment of SOH:

[0101] Measuring internal resistance: When using SPM simulation, by comparing the change in internal resistance during different charge and discharge cycles, the current battery internal resistance increment can be determined.

[0102] Calculate SOH: Calculate the battery's SOH based on the ratio of the current internal resistance (i.e., the internal resistance at the next moment predicted by the neural network) to the new battery reference internal resistance. For example, if the internal resistance increases by 20%, the SOH will decrease to 80% of the initial value.

[0103] Adjustment under aging state: When the internal resistance increases to a certain threshold (for example, when the internal resistance exceeds twice the initial value), the SOH value will drop sharply, which means that the battery has aged seriously and measures need to be taken such as reducing the charge and discharge rate or replacing the battery.

[0104] Among them, parameter adjustment in the SPM model: In the SPM model, the effect of internal resistance change on SOH can be adjusted and simulated in the following ways:

[0105] Internal resistance setting range: Under simulation conditions, the internal resistance range is set from 20mΩ to 150mΩ. This range covers the internal resistance changes from new batteries to severely aged batteries. By simulating the battery voltage response under different internal resistance values, the change of SOH can be observed.

[0106] Response under different aging conditions: The SPM model can simulate the physical and chemical behavior of the battery in detail, so it can output different voltage response curves under different aging conditions to reflect the impact of increased internal resistance.

[0107] Dynamically adjust the step size: In a simulation with a time step of 1 second and a total time of 3600 seconds, the change in internal resistance can be calculated regularly and the SOH value can be updated.

[0108] (2) Feedback the physical and chemical characteristics output by the SPM model to the LSTM model to update the weights and biases of the LSTM model. Through this feedback mechanism, the LSTM model can learn the aging characteristics of the battery more accurately and improve the prediction accuracy. This step is specifically divided into the following sub-steps:

[0109] (201) The voltage, current, internal resistance and other characteristic values ​​output by the SPM model are used as input data for a new round of LSTM model training;

[0110] (202) The LSTM model retrains its network weights and biases based on the input of the SPM model so that it can more accurately capture the aging trend of the battery;

[0111] (203) The LSTM model continuously adjusts its parameters through multiple rounds of iterative optimization to achieve the best prediction effect;

[0112] (204) Process details: After each feedback update of the LSTM weights, 50 rounds of fine-tuning training were performed with a learning rate of 0.0001; the mean square error (MSE) was used as the loss function to ensure stable convergence of the model.

[0113] If the V, I, and R input into the SPM are corrected, the corrected values ​​are returned to the LSTM. If not corrected, the original values ​​are returned.

[0114] The Adam optimizer is used to optimize the combined model to reduce the prediction error. The parameters of the LSTM model and the SPM model are optimized together to ensure the accuracy and stability of the model. The specific steps are as follows:

[0115] Define the loss function and calculate the error between the LSTM model prediction result and the actual observation value;

[0116] The Adam optimizer is used to adjust the learning rate and parameters of the LSTM model according to the loss function value; Adam optimizer parameter settings: initial learning rate: 0.001; β1 = 0.9, β2 = 0.999, ε = 1e-8;

[0117] The optimized LSTM model parameters are used as input to retrain the SPM model and adjust its parameters to reduce the overall prediction error.

[0118] SPM model details: The updated LSTM prediction value is used as the initial condition to re-simulate the SPM model. Compare the SPM model output with the actual observation value, adjust the battery physical and chemical parameters, and ensure the simulation accuracy.

[0119] Use the test set data to evaluate the model performance, adjust the model parameters, and ensure the generalization ability of the model. By verifying the test data, the prediction accuracy and robustness of the combined model are evaluated. The specific steps are as follows:

[0120] Select the data set that did not participate in the training as the test set to evaluate the combined model;

[0121] Calculate the prediction error on the test set to evaluate the accuracy and robustness of the model; prediction error evaluation indicators: mean square error (MSE): evaluates the difference between the predicted value and the true value; mean absolute error (MAE): evaluates the average absolute difference between the predicted value and the true value;

[0122] According to the evaluation results, the model parameters and optimization algorithms are adjusted to ensure the stability and accuracy of the combined model under different working conditions.

[0123] Step 4: Algorithm optimization and parallel computing.

[0124] Algorithms and models can be improved based on the following aspects:

[0125] Parallel computing: Utilize the parallel processing capabilities of cloud computing to speed up model training and reasoning and support large-scale data processing.

[0126] Adaptive model update: Implement an adaptive learning mechanism to dynamically adjust model parameters based on newly collected data to improve the real-time adaptability of the model.

[0127] Edge learning: Perform partial model training on the vehicle side to protect data privacy while reducing data transmission volume.

[0128] Data security: Implement data encryption and access control mechanisms to ensure the security of data transmission and storage between the vehicle and the cloud.

[0129] System redundancy: Build redundant systems and backup mechanisms to ensure that the system can still operate stably in unexpected situations.

[0130] Step 5: System integration and application.

[0131] The present embodiment provides a method for predicting the remaining cycle life of a battery, which can be specifically applied in three aspects: battery health management system, predictive maintenance and circular economy.

[0132] Battery health management system: Build a complete battery health management system, including data collection, processing, analysis and visualization, provide a user-friendly interface, establish a user feedback channel, and collect battery usage experience and problems. Provide personalized battery health reports, users can view battery status and usage suggestions, and improve user participation and satisfaction.

[0133] Predictive maintenance: Based on the RUL prediction results, predictive maintenance suggestions are provided to prevent battery failure and extend battery life.

[0134] Circular economy: Introduce battery recycling and reuse, and determine the secondary utilization value of recycled batteries through RUL assessment.

[0135] Through these detailed steps and improvements, the patent is not only more perfect in technology, but also more operational in practical applications, providing a set of advanced and reliable solutions for battery management systems.

[0136] Embodiment 2

[0137] This embodiment provides a device for predicting the remaining cycle life of a battery.

[0138] A battery remaining cycle life prediction device, comprising:

[0139] A data acquisition module is configured to: acquire battery status at several historical moments, wherein the battery status includes voltage, current, SOC, SOH and internal resistance;

[0140] The interactive prediction module is configured as follows: based on the battery status data of several historical moments, after predicting the battery status at the next moment through the neural network, the SPM model calculates the voltage, current and internal resistance inside the battery according to the predicted SOC and SOH; if the voltage, current and internal resistance calculated by the SPM model are consistent with those predicted by the neural network, the SPM model adjusts the battery physical and chemical parameters according to the battery status at the next moment predicted by the neural network; otherwise, based on the voltage, current and internal resistance calculated by the SPM model and the SOC and SOH predicted by the LSTM neural network as a set of training data, the neural network is fine-tuned and trained, and the battery status at the next moment is re-predicted;

[0141] The life calculation module is configured to calculate the remaining cycle life of the battery based on the battery state at the next moment predicted by the neural network.

[0142] The SPM model calculates the voltage inside the battery based on the predicted SOC. The specific steps include:

[0143] Based on the predicted SOC, the voltage is simulated through the corresponding relationship curve between voltage and SOC;

[0144] If the error between the simulated voltage and the voltage at the next moment predicted by the neural network exceeds the set value, the simulated voltage is used as the voltage output by the SPM model; otherwise, the voltage at the next moment predicted by the neural network is used as the voltage output by the SPM model.

[0145] The SPM model calculates the internal resistance of the battery based on the predicted SOH. The specific steps include:

[0146] Calculate the ratio of the next moment's internal resistance predicted by the neural network to the reference internal resistance, and simulate the battery's SOH;

[0147] If the error between the simulated SOH and the SOH at the next moment predicted by the neural network exceeds the set value, the internal resistance of the battery is calculated based on the SOH at the next moment predicted by the neural network and the reference internal resistance as the internal resistance output by the SPM model; otherwise, the internal resistance at the next moment predicted by the neural network is used as the internal resistance output by the SPM model.

[0148] Among them, the voltage and internal resistance at the historical moment are obtained by extracting features from the high-frequency pulse current.

[0149] It should be noted that the battery remaining cycle life prediction device provided in the above embodiment is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the electronic device is divided into different functional modules to complete all or part of the functions described above. In addition, the battery remaining cycle life calculation device provided in the above embodiment and the battery remaining cycle life calculation method embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0150] Embodiment 3

[0151] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the method for predicting the remaining cycle life of a battery as described in the first embodiment above are implemented.

[0152] Embodiment 4

[0153] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in a method for predicting remaining cycle life of a battery as described in the first embodiment above are implemented.

[0154] Figure 3 The schematic diagram of the structure of a computer device provided by an exemplary embodiment of the present application is shown. The computer device includes: a processor and a memory.

[0155] The processor may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field Programmable Gate Array), and PLA (Programmable Logic Array). The processor may also include a main processor and a coprocessor. The main processor is a processor for processing data in an awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in a standby state. In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content that needs to be displayed on the display screen. In some embodiments, the processor may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0156] The memory may include one or more computer-readable storage media, which may be non-transitory. The memory may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory is used to store at least one computer program, which is used to be used by the processor to implement the vehicle occupant detection method provided in the method embodiment of the present application.

[0157] Those skilled in the art will appreciate that the illustrated structure does not constitute a limitation on the computer device, and may include more or fewer components than illustrated, or combine certain components, or adopt a different component arrangement.

[0158] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting the remaining cycle life of a battery, characterized in that: include: Obtaining battery status at several historical moments, wherein the battery status includes voltage, current, SOC, SOH, and internal resistance; Based on the battery status data at several historical moments, the neural network is used to predict the battery status at the next moment. The SPM model calculates the voltage, current and internal resistance of the battery according to the predicted SOC and SOH. If the voltage, current and internal resistance calculated by the SPM model are consistent with those predicted by the neural network, the SPM model adjusts the battery physical and chemical parameters according to the battery state at the next moment predicted by the neural network; otherwise, the voltage, current and internal resistance calculated by the SPM model and the SOC and SOH predicted by the LSTM neural network are used as a set of training data, and the neural network is fine-tuned and trained to re-predict the battery state at the next moment; The remaining cycle life of the battery is calculated based on the battery status at the next moment predicted by the neural network.

2. A method for predicting remaining cycle life of a battery according to claim 1, characterized in that: The SPM model calculates the voltage inside the battery according to the predicted SOC, and the specific steps include: Based on the predicted SOC, the voltage is simulated through the corresponding relationship curve between voltage and SOC; If the error between the simulated voltage and the voltage at the next moment predicted by the neural network exceeds the set value, the simulated voltage is used as the voltage output by the SPM model; otherwise, the voltage at the next moment predicted by the neural network is used as the voltage output by the SPM model.

3. A method for predicting remaining cycle life of a battery according to claim 1, characterized in that: The SPM model calculates the internal resistance of the battery according to the predicted SOH, and the specific steps include: Calculate the ratio of the next moment's internal resistance predicted by the neural network to the reference internal resistance, and simulate the battery's SOH; If the error between the simulated SOH and the SOH at the next moment predicted by the neural network exceeds the set value, the internal resistance of the battery is calculated based on the SOH at the next moment predicted by the neural network and the reference internal resistance as the internal resistance output by the SPM model; otherwise, the internal resistance at the next moment predicted by the neural network is used as the internal resistance output by the SPM model.

4. A method for predicting remaining cycle life of a battery according to claim 1, characterized in that: The voltage and internal resistance at the historical moment are obtained by extracting features from the high-frequency pulse current.

5. A battery remaining cycle life prediction device, characterized in that: include: A data acquisition module is configured to: acquire battery status at several historical moments, wherein the battery status includes voltage, current, SOC, SOH and internal resistance; The interactive prediction module is configured as follows: based on the battery status data of several historical moments, after predicting the battery status at the next moment through a neural network, the SPM model calculates the voltage, current and internal resistance of the battery according to the predicted SOC and SOH; If the voltage, current and internal resistance calculated by the SPM model are consistent with those predicted by the neural network, the SPM model adjusts the battery physical and chemical parameters according to the battery state at the next moment predicted by the neural network; otherwise, the voltage, current and internal resistance calculated by the SPM model and the SOC and SOH predicted by the LSTM neural network are used as a set of training data, and the neural network is fine-tuned and trained to re-predict the battery state at the next moment; The life calculation module is configured to calculate the remaining cycle life of the battery based on the battery state at the next moment predicted by the neural network.

6. A battery remaining cycle life prediction device according to claim 5, characterized in that: The SPM model calculates the voltage inside the battery according to the predicted SOC, and the specific steps include: Based on the predicted SOC, the voltage is simulated through the corresponding relationship curve between voltage and SOC; If the error between the simulated voltage and the voltage at the next moment predicted by the neural network exceeds the set value, the simulated voltage is used as the voltage output by the SPM model; otherwise, the voltage at the next moment predicted by the neural network is used as the voltage output by the SPM model.

7. A battery remaining cycle life prediction device according to claim 5, characterized in that: The SPM model calculates the internal resistance of the battery according to the predicted SOH, and the specific steps include: Calculate the ratio of the next moment's internal resistance predicted by the neural network to the reference internal resistance, and simulate the battery's SOH; If the error between the simulated SOH and the SOH at the next moment predicted by the neural network exceeds the set value, the internal resistance of the battery is calculated based on the SOH at the next moment predicted by the neural network and the reference internal resistance as the internal resistance output by the SPM model; otherwise, the internal resistance at the next moment predicted by the neural network is used as the internal resistance output by the SPM model.

8. A battery remaining cycle life prediction device according to claim 5, characterized in that: The voltage and internal resistance at the historical moment are obtained by extracting features from the high-frequency pulse current.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in a method for predicting remaining cycle life of a battery as described in any one of claims 1 to 4 are implemented.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the method for predicting the remaining cycle life of a battery as described in any one of claims 1-4 are implemented.

Images