Power battery health management method and device, vehicle and storage medium

Abstract

This application relates to a method, device, vehicle, and storage medium for the health management of power batteries. The method includes: acquiring the current operating state parameters of any power battery in an on-board power battery system; performing forward extrapolation on the current twin model of any power battery to obtain the number of extrapolation cycles required for capacity degradation to reach a preset retirement threshold; calculating residual features by combining the current operating state parameters and the predicted operating state parameters output by the current twin model; determining the health state information of any power battery by combining the number of extrapolation cycles, the residual features, and the current operating state parameters; and executing battery health management actions matching the health state information. This solves the problems in related technologies, such as the difficulty of pre-trained static models accurately depicting the actual degradation behavior of batteries in the mid-to-late stages, resulting in poor prediction accuracy, and the inability of general models to depict the personalized degradation path of individual batteries, thus failing to achieve refined health management of individual batteries and affecting vehicle range.

Description

Technical Field

[0001] This application relates to the field of digital data processing technology, and in particular to a method, device, vehicle, and storage medium for health management of power batteries. Background Technology

[0002] In related technologies, battery life can be estimated by constructing a power battery model, and then the maintenance and management of the power battery can be carried out based on the estimation results. However, under long-term charge-discharge cycles, temperature fluctuations, and operating condition disturbances, the internal electrochemical characteristics, internal resistance decay, and capacity aging law of power batteries will undergo significant nonlinear changes. Static models pre-trained at the factory cannot age synchronously with the physical battery, making it difficult to accurately depict the actual degradation behavior of the battery in the middle and late stages. This causes the accuracy of remaining life estimation and state perception to rapidly decline with the use time. Moreover, the model is a generalized model based on a large number of samples, which only learns the average degradation law of a large number of batteries and ignores the different individual aging trajectories of different power batteries caused by differences in user charging habits, driving area environment, and load conditions. The generalized model cannot depict the personalized degradation path of a single battery, and it is even more difficult to capture the specific early fault characteristics such as micro-short circuits and local aging of a single cell. It can only output generalized assessment results and cannot achieve refined health assessment and personalized management of a single battery, which urgently needs improvement. Summary of the Invention

[0003] This application provides a method, device, vehicle, and storage medium for the health management of a power battery, in order to solve the technical problems in the related technologies, such as the inability of pre-trained static models to age synchronously with the physical battery, making it difficult to accurately depict the actual degradation behavior of the battery in the middle and late stages, resulting in poor prediction accuracy, and the inability of general models to depict the personalized degradation path of a single battery, thus failing to achieve refined health management of a single battery and affecting the vehicle's range.

[0004] The first aspect of this application provides a health management method for a power battery, comprising the following steps: obtaining the current operating state parameters of any power battery in an on-board power battery system, and determining a current twin model corresponding to any power battery; extrapolating the current twin model to obtain the number of extrapolation cycles when the capacity of any power battery decays to a preset retirement threshold; obtaining the predicted operating state parameters output by the current twin model, and using the predicted operating state parameters and the current operating state parameters to calculate the residual characteristics of any power battery; combining the number of extrapolation cycles, the residual characteristics, and the current operating state parameters to determine the health state information of any power battery, and performing battery health management actions matching the health state information.

[0005] Optionally, in one embodiment of this application, before obtaining the current operating state parameters of any power battery in the vehicle power battery system and determining the current twin model corresponding to any power battery, the method further includes: obtaining the initial operating state parameters of multiple power batteries in the vehicle power battery system within a preset acquisition period; constructing an initial twin model of the corresponding power battery based on the initial operating state parameters; obtaining the first cycle operating state parameters of the multiple power batteries within the first preset time period every first preset time interval; iteratively updating the initial twin model using the first cycle operating state parameters until the current twin model corresponding to the multiple power batteries is obtained.

[0006] Optionally, in one embodiment of this application, after obtaining the number of simulation cycles when the capacity of any of the power batteries decays to a preset retirement threshold, the method further includes: adjusting the first preset duration based on the number of simulation cycles to obtain a second preset duration; and every second preset duration, obtaining second-cycle operating state parameters of multiple power batteries within the second preset duration period, so as to update the current twin model using the second-cycle operating state parameters.

[0007] Optionally, in one embodiment of this application, the step of forward extrapolating the current twin model to obtain the number of extrapolation cycles when the capacity of any of the power batteries decays to a preset retirement threshold includes: acquiring historical operating condition data of any of the power batteries; obtaining multiple operating conditions of any of the power batteries based on the historical operating condition data; using the multiple operating conditions to forward extrapolate the current twin model to obtain multiple initial extrapolation cycle numbers; obtaining a probability distribution of the number of extrapolation cycles of any of the power batteries based on the multiple initial extrapolation cycle numbers; and obtaining the number of extrapolation cycles based on the probability distribution.

[0008] Optionally, in one embodiment of this application, the step of forward extrapolating the current twin model to obtain the number of extrapolation cycles when the capacity of any of the power batteries decays to a preset retirement threshold includes: acquiring historical usage condition data of any of the power batteries; based on the historical usage condition data, calculating the usage probability of multiple usage conditions and estimating the future duration corresponding to the multiple usage conditions; using the multiple usage conditions, forward extrapolating the current twin model to obtain a plurality of initial extrapolation cycle numbers; and based on the usage probability and the future duration, performing a weighted summation of the plurality of initial extrapolation cycle numbers to obtain the total number of extrapolation cycles.

[0009] Optionally, in one embodiment of this application, the step of calculating the residual characteristics of any of the power batteries using the predicted operating state parameters and the current operating state parameters includes: obtaining a real data sequence of any of the power batteries sorted by acquisition time based on the current operating state parameters; obtaining a predicted data sequence of any of the power batteries sorted by acquisition time based on the predicted operating state parameters; calculating a residual sequence by combining the real data sequence and the predicted data sequence; and calculating the root mean square error, residual kurtosis, and residual spectral dominant frequency energy of the residual characteristics based on the residual sequence.

[0010] A second aspect of this application provides a health management device for a power battery, comprising: an acquisition module for acquiring current operating state parameters of any power battery in an on-board power battery system and determining a current twin model corresponding to any power battery; a deduction module for deducing the current twin model forward to obtain the number of deduction cycles when the capacity of any power battery decays to a preset retirement threshold; a prediction module for acquiring predicted operating state parameters output by the current twin model, and using the predicted operating state parameters and the current operating state parameters to calculate residual characteristics of any power battery; and a control module for combining the number of deduction cycles, the residual characteristics, and the current operating state parameters to determine health state information of any power battery, and executing battery health management actions matching the health state information.

[0011] Optionally, in one embodiment of this application, it further includes: a second acquisition module, configured to acquire initial operating state parameters of multiple power batteries in the vehicle power battery system within a preset acquisition period; a construction module, configured to construct an initial twin model of the corresponding power battery based on the initial operating state parameters; a third acquisition module, configured to acquire first-cycle operating state parameters of the multiple power batteries within the first preset time period every first preset time interval; and a first update module, configured to iteratively update the initial twin model using the first-cycle operating state parameters until the current twin model corresponding to the multiple power batteries is obtained.

[0012] Optionally, in one embodiment of this application, it further includes: an adjustment module, configured to adjust the first preset duration based on the number of simulation cycles to obtain a second preset duration; and a second update module, configured to acquire multiple second-cycle operating state parameters of the power battery within the second preset duration period every second preset duration, so as to update the current twin model using the second-cycle operating state parameters.

[0013] Optionally, in one embodiment of this application, the deduction module includes: a first acquisition unit, configured to acquire historical operating condition data of any of the power batteries; a second acquisition unit, configured to obtain multiple operating conditions of any of the power batteries based on the historical operating condition data; a first deduction unit, configured to use the multiple operating conditions to deduce the current twin model backwards respectively, and obtain a corresponding multiple initial deduction cycle counts; and a first calculation unit, configured to obtain a probability distribution of the deduction cycle count of any of the power batteries based on the multiple initial deduction cycle counts, and obtain the deduction cycle count based on the probability distribution.

[0014] Optionally, in one embodiment of this application, the extrapolation module includes: a third acquisition unit, configured to acquire historical usage condition data of any of the power batteries; a second calculation unit, configured to calculate the usage probability of multiple usage conditions based on the historical usage condition data, and estimate the future duration corresponding to the multiple usage conditions; a second extrapolation unit, configured to extrapolate the current twin model forward using the multiple usage conditions to obtain a plurality of initial extrapolation cycle counts; and a third calculation unit, configured to perform a weighted summation of the plurality of initial extrapolation cycle counts based on the usage probability and the future duration to obtain the extrapolation cycle count.

[0015] Optionally, in one embodiment of this application, the prediction module includes: a fourth acquisition unit, configured to obtain a real data sequence of any of the power batteries sorted by acquisition time based on the current operating state parameters; a fifth acquisition unit, configured to obtain a predicted data sequence of any of the power batteries sorted by acquisition time based on the predicted operating state parameters; a fourth calculation unit, configured to calculate a residual sequence by combining the real data sequence and the predicted data sequence; and a fifth calculation unit, configured to calculate the root mean square error, residual kurtosis, and residual spectral dominant frequency energy of the residual features based on the residual sequence.

[0016] A third aspect of this application provides a vehicle, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the power battery health management method as described in the above embodiments.

[0017] A fourth aspect of this application provides a computer-readable storage medium storing computer instructions for causing the computer to perform the power battery health management method as described in the above embodiments.

[0018] A fifth aspect of this application provides a computer program product, including a computer program, which, when executed, is used to implement the above-described health management method for a power battery.

[0019] This application embodiment can construct a corresponding twin model for each individual power battery in an on-board power battery system. The twin model is used to predict the number of cycles required for the capacity of a single power battery to decay to a retirement threshold. Residual features are obtained by comparing the predicted data output by the twin model with actual data, capturing early anomalies in individual batteries. Combining the predicted cycle count, residual features, and current operating state parameters, the health status information of the individual power battery is determined, and battery health management actions matching the health status information are executed, improving vehicle safety and the economic value of the battery throughout its entire lifecycle. This solves the technical problems in related technologies where pre-trained static models cannot age synchronously with the physical battery, making it difficult to accurately depict the actual degradation behavior of the battery in the later stages, resulting in poor prediction accuracy. Furthermore, general-purpose models cannot depict the personalized degradation path of individual batteries, failing to achieve refined health management of individual batteries, thus affecting vehicle range.

[0020] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0021] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a health management method for a power battery according to an embodiment of this application; Figure 2 This is a schematic diagram illustrating the principle of a health management method for a power battery according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a health management device for a power battery according to an embodiment of this application; Figure 4 This is a structural schematic diagram of a vehicle provided according to an embodiment of this application. Detailed Implementation

[0022] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0023] The following description, with reference to the accompanying drawings, outlines a method, apparatus, vehicle, and storage medium for managing the health of a power battery according to embodiments of this application. Addressing the technical problems mentioned in the background section, such as the inability of pre-trained static models to synchronize with the aging of physical batteries, difficulty in accurately depicting the actual degradation behavior of batteries in the later stages, poor prediction accuracy, and the inability of general models to depict the individual degradation paths of single batteries, thus hindering refined health management of individual batteries and affecting vehicle range, this application provides a method for managing the health of a power battery. In this method, a corresponding twin model can be constructed for each individual power battery in an on-board power battery system. The twin model is used to deduce the number of simulation cycles required for the capacity of a single power battery to decay to a retirement threshold. Residual features are obtained by comparing the predicted data output by the twin model with the actual data, capturing early anomalies in individual batteries. Combining the number of simulation cycles, residual features, and current operating state parameters, the health status information of the individual power battery is determined, and battery health management actions matching the health status information are executed, thereby improving vehicle safety and the economic value of the battery throughout its entire lifecycle. This solves the technical problems in related technologies, such as the inability of pre-trained static models to age synchronously with physical batteries, making it difficult to accurately depict the actual degradation behavior of batteries in the mid-to-late stages, resulting in poor prediction accuracy, and the inability of general models to depict the personalized degradation path of individual batteries, thus failing to achieve refined health management of individual batteries and affecting vehicle range.

[0024] Specifically, Figure 1 This is a schematic flowchart illustrating a health management method for a power battery provided in an embodiment of this application.

[0025] like Figure 1 As shown, the health management method for this power battery includes the following steps: In step S101, the current operating state parameters of any power battery in the vehicle power battery system are obtained, and the current twin model corresponding to any power battery is determined.

[0026] In actual operation, the vehicle power battery system can continuously collect data to obtain data related to the operating status of a power battery when it is necessary to confirm the health status of a power battery. The data types of the data continuously collected by the vehicle power battery system can be shown in Table 1.

[0027] Table 1

[0028] After data collection, embodiments of this application can also clean the data to remove outliers such as sensor noise and communication packet loss. To reduce communication load, embodiments of this application can also employ a strategy of combining change-triggered data with key event packaging for data upload. For example, during charging, only when a characteristic peak appears on the dQ / dV curve is the high sampling rate data uploaded; during rest periods, only the first 30 minutes of data from the voltage relaxation curve (containing the main relaxation information) is uploaded; under normal driving conditions, the compressed feature vector is uploaded at a frequency of 1Hz.

[0029] Of the collected current operating status data, a portion can be used to form a real data sequence for subsequent residual feature calculation; another portion can serve as the data basis for power battery health management; and a third portion can be used for data processing, feature extraction, and subsequent incremental learning of the power battery's twin model. The data basis for residual feature calculation and power battery health management will be described later; here, we describe the data processing.

[0030] Taking data uploading to the cloud as an example, this application embodiment can first assign a unique digital identity to each power battery, and then archive the data of each power battery collected by the vehicle power battery system according to the digital identity, thereby drawing a corresponding individual data profile for each power battery.

[0031] In order to deeply characterize the internal state of the power battery, this embodiment of the application can calculate the following key features in real time by collecting the current operating state parameters, and then use the key features to perform incremental learning of the twin model. The incremental learning of the twin model will be described later.

[0032] Characteristics of the capacity increment curve (dQ / dV): Voltage during charging With capacity The relationship, calculation:

[0033] After smoothing and filtering the voltage range, the position and height of the characteristic peaks are extracted as an important fingerprint of the battery aging state.

[0034] Characteristics of the resting voltage relaxation curve fitting: During the resting phase after charging is complete, the voltage... The variation with time t can be described by a double exponential model:

[0035] Parameters are obtained through nonlinear least squares fitting. , , , These parameters are sensitive to the internal polarization state.

[0036] AC internal resistance characteristics: At a specific frequency Below, by injecting a small-amplitude sinusoidal current And measure the response voltage Calculate the impedance magnitude:

[0037] in , For amplitude, For phase difference, Impedance spectra at different frequencies can be used for early warning of lithium plating risks.

[0038] This application embodiment can also pre-draw an individual data profile for each power battery in the vehicle power battery system, that is, build a corresponding twin model for each power battery. The twin model is constructed by using the operating status parameters during the first 100 complete charge-discharge cycles (or the first 30-day operating cycle) after the power battery is connected to the vehicle power battery system. The twin model is then periodically self-evolved as the power battery is used. Each power battery and its corresponding twin model can also be associated through a digital identity. When it is necessary to confirm the health status of a power battery, the corresponding twin model can be retrieved through the digital identity for subsequent health status confirmation.

[0039] Optionally, in one embodiment of this application, before obtaining the current operating state parameters of any power battery in the vehicle power battery system and determining the current twin model corresponding to any power battery, the method further includes: obtaining the initial operating state parameters of multiple power batteries in the vehicle power battery system within a preset acquisition period; constructing an initial twin model of the corresponding power battery based on the initial operating state parameters; obtaining the first cycle operating state parameters of the multiple power batteries within a first preset time period every first preset time interval; iteratively updating the initial twin model using the first cycle operating state parameters until the current twin model corresponding to the multiple power batteries is obtained.

[0040] In actual implementation, the embodiments of this application can automatically generate the corresponding initial twin model after the first 100 complete charge-discharge cycles (or the first 30-day operating cycle) after any power battery is connected to the vehicle power battery system.

[0041] When constructing the initial twin model, embodiments of this application may use a gated recurrent unit (GRU)-attention mechanism hybrid neural network as the basic model architecture. The gated recurrent unit is good at processing time series data and capturing the long-term dependence of battery degradation; the attention mechanism can automatically identify the influence weight of different features (such as temperature and charge / discharge rate) on the health status.

[0042] Based on the basic model architecture, this application embodiment can use historical data of the power battery itself during the initialization cycle, such as the charging capacity, average temperature, charge / discharge rate, and characteristic peak position of the dQ / dV curve for each cycle, as model input; and use the power battery health status value estimated by the on-board power battery system at the end of each cycle (as a relative benchmark) as output to generate an initial model parameter set as training data to anchor the original chemical characteristics and micro-manufacturing differences of the power battery.

[0043] After obtaining the initial twin model, the embodiments of this application can perform self-evolution of the twin model.

[0044] Each time the vehicle completes a full charge or accumulates 7 days of operation, online incremental learning can be automatically triggered, enabling the model to accurately capture the unique capacity decay inflection point and internal resistance leap step of this battery. As the battery ages, the model's representation accuracy of the current state remains at a high confidence level.

[0045] For example, embodiments of this application can extract the latest data generated by the power battery since the last update to construct a new training sample set. Starting with the current model parameters, the model is trained iteratively using only the new data in a lightweight manner. The Elastic Weight Consolidation (EWC) algorithm is employed to protect key network layer parameters that characterize the battery's historical properties while updating the model to adapt to the new state, preventing catastrophic forgetting. The EWC loss function is used during model fine-tuning to maintain a balance between new knowledge learning and old knowledge.

[0046] in, The prediction loss (e.g., mean squared error) for the current batch of data. This is the first update since the last update. One network parameter; These are the diagonal elements of the Fisher information matrix, measuring the importance of parameters to historical tasks; To balance the factors, the intensity of historical knowledge retention is controlled.

[0047] This mechanism can effectively prevent catastrophic forgetting, enabling the model to retain memories of early aging characteristics while continuously learning new states.

[0048] Furthermore, if the model is saved to the cloud in this embodiment of the application, the cloud can save the model version number and the corresponding update timestamp after each update, which is convenient for backtracking and comparative analysis.

[0049] In addition, to reduce computational difficulty, embodiments of this application can also cluster multiple power batteries in the vehicle power battery system, such as clustering based on location or clustering based on the similarity of operating state parameters, and calculate the mean for multiple power batteries in each cluster, thereby constructing a twin model corresponding to each cluster for subsequent health management.

[0050] In step S102, the current twin model is extrapolated forward to obtain the number of extrapolation cycles when the capacity of any power battery decays to a preset retirement threshold.

[0051] As one possible approach, embodiments of this application can utilize the current twin model obtained after self-learning updates to predict the remaining lifespan of the power battery.

[0052] In this application embodiment, the future operating conditions of the power battery (such as one fast charge per day and an average temperature of 25°C) can be input, and the number of cycles required for the battery capacity to decay to the retirement threshold (such as 80%) can be simulated by extrapolating from the model.

[0053] Optionally, in one embodiment of this application, the process of forward extrapolating the current twin model to obtain the number of extrapolation cycles when the capacity of any power battery decays to a preset retirement threshold includes: acquiring historical operating condition data of any power battery; obtaining multiple operating conditions of any power battery based on the historical operating condition data; using the multiple operating conditions to forward extrapolate the current twin model to obtain multiple initial extrapolation cycle numbers; obtaining the probability distribution of the number of extrapolation cycles of any power battery based on the multiple initial extrapolation cycle numbers; and obtaining the number of extrapolation cycles based on the probability distribution.

[0054] In some embodiments, historical charge and discharge data of the individual power battery throughout its entire life cycle can be collected, including real-world usage information such as charging rate, operating temperature, depth of discharge, and charge / discharge duration. This historical data can be used to accurately reflect the battery's unique user habits and environmental characteristics, serving as the data foundation for subsequently constructing future operating conditions and conducting simulations.

[0055] Based on historical usage data, the embodiments of this application can summarize and extract the high-frequency and representative future usage conditions of the power battery from various historical usage conditions (such as fast charging high temperature conditions, slow charging normal temperature conditions, and low temperature low current conditions).

[0056] Furthermore, embodiments of this application can utilize various potential future operating conditions to retrospectively extrapolate the current twin model, independently simulating each operating condition to simulate the gradual capacity degradation of the power battery under that condition until a preset retirement threshold (e.g., 80% capacity) is reached. The remaining cycle count for each operating condition is then calculated, representing multiple initial extrapolation cycle counts. This process is equivalent to Monte Carlo simulation performing multiple sampling simulations of the operating condition random variables.

[0057] Among them, Monte Carlo simulation is used to consider the uncertainty of future operating conditions and output the probability distribution of remaining useful life (RUL) instead of a single value, providing a confidence interval for the health management of power batteries.

[0058] Let the capacity decay trajectory predicted by the model be... ,in, x For future operating condition parameters (charging rate, average temperature, etc.). Considering the uncertainties of the operating conditions, x Treating them as random variables, perform M sampling simulations:

[0059] The probability distribution of remaining useful life (RUL) can be obtained. On the one hand, multiple initial extrapolation cycles can be weighted and summed based on the probability of use under the operating condition and the future duration to obtain a single extrapolation cycle for health status determination. On the other hand, the confidence interval of the remaining useful life can be output, with the 5th percentile as a conservative estimate and the 95th percentile as an optimistic estimate, so as to achieve a life prediction that takes into account both deterministic output and uncertainty risk assessment.

[0060] Optionally, in one embodiment of this application, the process of forward extrapolating the current twin model to obtain the number of extrapolation cycles when the capacity of any power battery decays to a preset retirement threshold includes: acquiring historical usage data of any power battery; based on the historical usage data, calculating the usage probability of various usage conditions and estimating the future duration corresponding to various usage conditions; using various usage conditions to forward extrapolate the current twin model to obtain multiple initial extrapolation cycle numbers; and weighting and summing the multiple initial extrapolation cycle numbers based on the usage probability and the future duration to obtain the total number of extrapolation cycles.

[0061] In other embodiments, based on historical usage data (such as temperature, rate, depth of charge and discharge, and driving conditions), the probability of use of the power battery under each usage condition in the past, as well as the possible duration of future use, can be calculated.

[0062] For each usage condition, the embodiments of this application can use a twin model to extrapolate forward and calculate the number of cycles at which the battery will fail under each condition. Then, by combining the usage probability and future duration, the lifespan results of each condition are weighted and averaged to obtain the final total number of usable cycles.

[0063] This application embodiment uses the historical usage conditions of the power battery to statistically analyze the probability of the conditions and the future duration, and conducts digital twin life simulations for different typical conditions. Then, by weighting and fusing the simulation results of multiple conditions based on probability and duration, it accurately obtains the remaining number of cycles before the battery capacity decays to the retirement threshold, which is more in line with the actual driving habits of users.

[0064] Optionally, in one embodiment of this application, after obtaining the number of simulation cycles when the capacity of any power battery decays to a preset retirement threshold, the method further includes: adjusting a first preset duration based on the number of simulation cycles to obtain a second preset duration; and every second preset duration, acquiring second-cycle operating state parameters of multiple power batteries within the second preset duration period to update the current twin model using the second-cycle operating state parameters.

[0065] It is understandable that the degradation of a power battery is not a simple linear relationship with its usage time. The shorter the remaining lifespan, the more likely the power battery has entered a relatively aging stage, and the faster its performance will degrade and the more frequent its state changes. The longer the remaining lifespan, the more likely the power battery is in a healthy state, with smooth performance degradation and stable state changes.

[0066] Based on this, the embodiments of this application can adjust the data acquisition frequency for model self-updating according to the number of simulation cycles. For example, when the number of simulation cycles is 80, the twin model is updated every seven days; when the number of simulation cycles is 50, the twin model is updated every five days, to ensure that the life prediction and fault diagnosis of the power battery before its retirement are still accurate.

[0067] In step S103, the predicted operating state parameters output by the current twin model are obtained, and the residual characteristics of any power battery are calculated using the predicted operating state parameters and the current operating state parameters.

[0068] Furthermore, embodiments of this application can employ virtual-real residual analysis technology. The actual sensor data of the power battery at the previous data acquisition moment is input into the twin model to obtain the predicted operating state parameters (such as the expected voltage curve) reconstructed by the model. The residual sequence between the current operating state parameters and the predicted operating state parameters is calculated, and the statistical characteristics of the residuals (such as root mean square error and frequency domain characteristics of the residual sequence) are extracted. Using the current operating state parameters as input, the twin model outputs new predicted operating state parameters, which are then compared with the operating state data acquired at the next data acquisition moment to calculate residual characteristics, thereby achieving continuous health management of the power battery.

[0069] Optionally, in one embodiment of this application, the residual characteristics of any power battery are calculated using predicted operating state parameters and current operating state parameters, including: obtaining a real data sequence of any power battery sorted by acquisition time based on the current operating state parameters; obtaining a predicted data sequence of any power battery sorted by acquisition time based on the predicted operating state parameters; calculating a residual sequence by combining the real data sequence and the predicted data sequence; and calculating the root mean square error, residual kurtosis, and residual spectral dominant frequency energy of the residual characteristics based on the residual sequence.

[0070] In actual implementation, the embodiments of this application can first define the residual sequence. ,in:

[0071] For the first The actual voltage (or current) at each sampling point. The predicted values ​​given by the twin model, To avoid dividing by zero for small constants.

[0072] The embodiments of this application can extract the statistical feature vector of the residuals: Root mean square error ; residual kurtosis ; Residual spectrum dominant frequency energy (Obtained via Fast Fourier Transform) The above features are input into an offline-trained Isolation Forest or Support Vector Machine classifier, when... Exceeding the dynamic threshold and When an abnormal peak value appears in a specific frequency band, it is determined to be an early fault.

[0073] In this embodiment, residual features can be input into a dedicated fault classifier (trained based on support vector machines or isolated forests). When the residual features exhibit a specific pattern (such as an abnormal increase in energy in a certain frequency band), it is determined to be a sign of early faults such as internal short circuits or lithium plating.

[0074] In step S104, the health status information of any power battery is determined by combining the number of simulation cycles, residual characteristics and current operating status parameters, and battery health management actions that match the health status information are executed.

[0075] The embodiments of this application can dynamically generate a fully personalized optimization strategy based on the number of simulation cycles, early fault symptoms determined based on residual characteristics, and battery health status (the ratio of the current available capacity of the power battery to the nominal capacity obtained based on the current operating state parameters).

[0076] For example, for power batteries with rapid capacity decay, a gentle charging curve is generated, the current in the constant current stage is appropriately reduced, and the constant voltage stage is entered earlier to reduce the risk of lithium plating; for batteries in good health, a fast charging curve is used.

[0077] For power batteries with significantly increased internal resistance, the thermal management system activation threshold is lowered from 45℃ to 42℃ to intervene in cooling earlier and delay the degradation caused by high temperature.

[0078] When it is predicted that the battery health of a certain power battery will drop below 80% within 3 months, maintenance suggestions or replacement reminders will be sent to users in advance.

[0079] This application embodiment can encapsulate the optimization strategy into a standard instruction package and send it to the vehicle power battery system. After parsing the instruction package, the vehicle power battery system executes control according to the new strategy during subsequent charging and discharging processes.

[0080] It is understandable that the above processes can be achieved by integrating high-performance computing units on the vehicle side, or by utilizing cloud computing power for data storage and decision-making, thereby enabling vehicle-side resources to be used for other vehicle functions.

[0081] like Figure 2 As shown, taking cloud storage data and decision-making as an example, the power battery health management method of this application embodiment can be applied to the vehicle and the cloud in actual execution, and realize the health management of the power battery through data exchange between the vehicle and the cloud.

[0082] The vehicle-side components can be used for data collection and health management, while the cloud-based components can be used for data storage, model updates, and decision-making.

[0083] Specifically, such as Figure 2 As shown, the following steps may be included: Step S1: Utilize the vehicle's power battery system to continuously collect high-precision data on individual cell voltage / total voltage (10Hz), charge / discharge capacity (10Hz), battery temperature / ambient temperature, and driving conditions / charging behavior.

[0084] Step S2 involves preprocessing the raw operating data collected from the vehicle power battery system to extract the feature data required for model training and state perception, including the characteristic peaks of the dQ / dV curve, the resting voltage relaxation curve, and the AC internal resistance at different frequencies. Simultaneously, data cleaning, anomaly removal, and compression packaging are completed.

[0085] Step S3: Upload the processed feature data and the original operating data to the cloud. The cloud receives and stores the data. By pre-configuring a unique digital identity for each vehicle power battery, the received original data, feature data and the corresponding power battery's digital identity are bound together, achieving accurate collection and retrieval of individual-level data.

[0086] In step S4, the cloud can initialize the twin model by selecting the self-operation data of the power battery within the preset initialization period (such as the first 100 charge-discharge cycles) to build training samples. The initial model architecture is built based on the GRU-attention mechanism network. The initial individual digital twin model is obtained by training the sample data, so that the twin model can accurately anchor the original chemical characteristics and micro-manufacturing differences of the corresponding single power battery from the initialization stage, and establish a high-fidelity digital foundation for accurate degradation trajectory tracking throughout the entire life cycle.

[0087] Based on this, the cloud performs incremental learning and online self-evolution updates periodically according to a preset time interval. It uses real-time operating data of the power battery to perform lightweight iterative training on the initial individual digital twin model. An elastic weight consolidation algorithm is introduced to protect key network parameters that characterize the battery's historical characteristics while updating the model to adapt to the current aging state of the battery, thus avoiding catastrophic amnesia. Through continuous online self-evolution updates, the individual digital twin model can adaptively track the individual-specific aging trajectory of the corresponding power battery, accurately capturing the battery's unique capacity decay inflection point, internal resistance jump step, and other personalized degradation characteristics.

[0088] In step S5, the cloud can perform forward extrapolation on the online self-evolving updated individual twin model based on the historical operating conditions of the power battery. Monte Carlo simulation can be used to consider the uncertainty of future operating conditions to obtain the number of extrapolation cycles for the power battery. The predicted operating state parameters of the power battery are reconstructed using the individual twin model, and virtual-real residual analysis is performed on them with the current real sensing data of the power battery to extract the statistical characteristics of the residual sequence for early fault diagnosis. The health status of the power battery is calculated by combining the capacity information output by the individual digital twin model. The remaining value assessment of the power battery throughout its entire life cycle is completed by comprehensively considering the number of extrapolation cycles, residual statistical characteristics, and battery health status. Based on the assessment results, a personalized charging strategy or thermal management strategy matching the current health status of the power battery is generated and converted into health management instructions and sent to the vehicle power battery system.

[0089] In step S6, the vehicle can receive health management instructions and complete the corresponding health management actions.

[0090] Furthermore, embodiments of this application can also obtain the execution effect of health management actions, and update the twin model based on the operating status parameters of the power battery fed back by the execution effect.

[0091] The power battery health management method proposed in this application can construct a corresponding twin model for each individual power battery in a vehicle power battery system. The twin model is used to predict the number of cycles required for the capacity of a single power battery to decay to a retirement threshold. Residual features are obtained by comparing the predicted data output by the twin model with actual data, capturing early anomalies in individual batteries. Combining the predicted cycle count, residual features, and current operating state parameters, the health status information of the individual power battery is determined, and battery health management actions matching the health status information are executed, improving vehicle safety and the economic value of the battery throughout its entire lifecycle. This solves the technical problems in related technologies where pre-trained static models cannot age synchronously with the physical battery, making it difficult to accurately depict the actual degradation behavior of the battery in the later stages, resulting in poor prediction accuracy. Furthermore, general models for the entire battery group cannot depict the personalized degradation path of a single battery, failing to achieve refined health management of individual batteries, thus affecting vehicle range.

[0092] Next, the health management device for a power battery according to an embodiment of this application is described with reference to the accompanying drawings.

[0093] Figure 3 This is a block diagram of a power battery health management device according to an embodiment of this application.

[0094] like Figure 3 As shown, the health management device 10 for the power battery includes: Specifically, the acquisition module 100 is used to acquire the current operating status parameters of any power battery in the vehicle power battery system and determine the current twin model corresponding to any power battery.

[0095] The deduction module 200 is used to deduce the current twin model forward to obtain the number of deduction cycles when the capacity of any power battery decays to a preset retirement threshold.

[0096] The prediction module 300 is used to obtain the predicted operating state parameters output by the current twin model, so as to calculate the residual characteristics of any power battery using the predicted operating state parameters and the current operating state parameters.

[0097] The control module 400 is used to determine the health status information of any power battery by combining the number of simulation cycles, residual characteristics and current operating status parameters, and to execute battery health management actions that match the health status information.

[0098] Optionally, in one embodiment of this application, the health management device 10 for the power battery further includes: The second acquisition module is used to acquire the initial operating status parameters of multiple power batteries in the vehicle power battery system within a preset acquisition period.

[0099] The module is used to build an initial twin model of the corresponding power battery based on the initial operating state parameters.

[0100] The third acquisition module is used to acquire the first cycle operating status parameters of multiple power batteries within the first preset time period every first preset time period.

[0101] The first update module is used to iteratively update the initial twin model using the first cycle operating state parameters until the current twin model corresponding to multiple power batteries is obtained.

[0102] Optionally, in one embodiment of this application, the health management device 10 for the power battery further includes: The adjustment module is used to adjust the first preset duration based on the number of simulation cycles to obtain the second preset duration.

[0103] The second update module is used to acquire the second-cycle operating status parameters of multiple power batteries within the second preset time period every second preset time period, so as to update the current twin model using the second-cycle operating status parameters.

[0104] Optionally, in one embodiment of this application, the deduction module 200 includes: The first acquisition unit is used to acquire historical operating condition data of any power battery.

[0105] The second acquisition unit is used to obtain multiple operating conditions for any power battery based on historical operating condition data.

[0106] The first deduction unit is used to use various operating conditions to deduce the current twin model forward, and obtain the corresponding number of initial deduction cycles.

[0107] The first calculation unit is used to obtain the probability distribution of the number of simulation cycles for any power battery based on multiple initial simulation cycle counts, and to obtain the number of simulation cycles based on the probability distribution.

[0108] Optionally, in one embodiment of this application, the deduction module 200 includes: The third acquisition unit is used to acquire historical operating condition data of any power battery.

[0109] The second calculation unit is used to calculate the probability of use for various usage conditions based on historical usage data, and to estimate the future duration of various usage conditions.

[0110] The second deduction unit is used to use various operating conditions to deduce the current twin model forward, and obtain the corresponding number of initial deduction cycles.

[0111] The third calculation unit is used to perform a weighted summation of multiple initial simulation cycle counts based on the usage probability and future duration to obtain the simulation cycle count.

[0112] Optionally, in one embodiment of this application, the prediction module 300 includes: The fourth acquisition unit is used to obtain the real data sequence of any power battery based on the acquisition time sorting, based on the current operating status parameters.

[0113] The fifth acquisition unit is used to obtain a predicted data sequence of any power battery based on the acquisition time sorting, based on the predicted operating state parameters.

[0114] The fourth calculation unit is used to calculate the residual sequence by combining the real data sequence and the predicted data sequence.

[0115] The fifth calculation unit is used to calculate the root mean square error, residual kurtosis, and residual spectrum dominant frequency energy based on the residual sequence.

[0116] It should be noted that the foregoing explanation of the embodiment of the power battery health management method also applies to the power battery health management device of this embodiment, and will not be repeated here.

[0117] The power battery health management device proposed in this application can construct a corresponding twin model for each individual power battery in a vehicle power battery system. The twin model is used to predict the number of cycles required for the capacity of a single power battery to decay to a retirement threshold. Residual features are obtained by comparing the predicted data output by the twin model with actual data, capturing early anomalies in individual batteries. Combining the predicted cycle count, residual features, and current operating state parameters, the health status information of the individual power battery is determined, and battery health management actions matching the health status information are executed, improving vehicle safety and the economic value of the battery throughout its entire lifecycle. This solves the technical problems in related technologies where pre-trained static models cannot age synchronously with the physical battery, making it difficult to accurately depict the actual degradation behavior of the battery in the later stages, resulting in poor prediction accuracy. Furthermore, general models for the entire battery group cannot depict the personalized degradation path of a single battery, failing to achieve refined health management of individual batteries, thus affecting vehicle range.

[0118] Figure 4 A schematic diagram of the structure of a vehicle provided in an embodiment of this application. The vehicle may include: The memory 401, the processor 402, and the computer program stored on the memory 401 and capable of running on the processor 402.

[0119] When the processor 402 executes the program, it implements the power battery health management method provided in the above embodiments.

[0120] Furthermore, the vehicle also includes: Communication interface 403 is used for communication between memory 401 and processor 402.

[0121] The memory 401 is used to store computer programs that can run on the processor 402.

[0122] Memory 401 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0123] If the memory 401, processor 402, and communication interface 403 are implemented independently, then the communication interface 403, memory 401, and processor 402 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0124] Optionally, in a specific implementation, if the memory 401, processor 402, and communication interface 403 are integrated on a single chip, then the memory 401, processor 402, and communication interface 403 can communicate with each other through an internal interface.

[0125] Processor 402 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0126] This embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described health management method for power batteries.

[0127] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the power battery health management method provided in this embodiment of the invention.

[0128] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do 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 one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0129] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0130] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0131] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing 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 (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0132] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0133] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0134] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0135] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A health management method for a power battery, characterized in that, Includes the following steps: Obtain the current operating status parameters of any power battery in the vehicle power battery system, and determine the current twin model corresponding to any of the power batteries; By extrapolating the current twin model forward, the number of extrapolation cycles required to determine when the capacity of any of the power batteries decays to a preset retirement threshold is obtained; Obtain the predicted operating state parameters output by the current twin model, and use the predicted operating state parameters and the current operating state parameters to calculate the residual characteristics of any of the power batteries; By combining the number of simulation cycles, the residual characteristics, and the current operating status parameters, the health status information of any of the power batteries is determined, and battery health management actions matching the health status information are executed.

2. The method according to claim 1, characterized in that, Before obtaining the current operating state parameters of any power battery in the vehicle power battery system and determining the current twin model corresponding to any of the power batteries, the process further includes: Acquire the initial operating status parameters of multiple power batteries in the vehicle power battery system within a preset acquisition period; An initial twin model of the corresponding power battery is constructed based on the initial operating state parameters; Every first preset time interval, acquire multiple power battery first cycle operating status parameters within the first preset time interval; The initial twin model is iteratively updated using the operating state parameters of the first cycle until the current twin models corresponding to the multiple power batteries are obtained.

3. The method according to claim 2, characterized in that, After obtaining the number of simulation cycles required for the capacity of any of the aforementioned power batteries to decay to a preset retirement threshold, the method further includes: The first preset duration is adjusted based on the number of iterations to obtain the second preset duration; Every second preset time interval, a plurality of second-cycle operating state parameters of the power battery within the second preset time interval are acquired, so as to update the current twin model using the second-cycle operating state parameters.

4. The method according to claim 1, characterized in that, The forward extrapolation of the current twin model to obtain the number of extrapolation cycles when the capacity of any of the power batteries decays to a preset retirement threshold includes: Obtain historical operating condition data for any of the aforementioned power batteries; Based on the historical usage data, multiple usage conditions for any of the power batteries are obtained; By utilizing various operating conditions, the current twin model is extrapolated forward to obtain multiple initial extrapolation cycle counts. Based on the multiple initial simulation cycle counts, a probability distribution of the simulation cycle count for any of the power batteries is obtained, and the simulation cycle count is obtained based on the probability distribution.

5. The method according to claim 1, characterized in that, The forward extrapolation of the current twin model to obtain the number of extrapolation cycles when the capacity of any of the power batteries decays to a preset retirement threshold includes: Obtain historical operating condition data for any of the aforementioned power batteries; Based on the historical usage data, the usage probability of various usage conditions is statistically analyzed, and the future duration of the various usage conditions is estimated. By utilizing various operating conditions, the current twin model is extrapolated forward to obtain multiple initial extrapolation cycle counts. Based on the usage probability and the future duration, the number of initial simulation cycles is weighted and summed to obtain the number of simulation cycles.

6. The method according to claim 1, characterized in that, The step of calculating the residual characteristics of any of the power batteries using the predicted operating state parameters and the current operating state parameters includes: Based on the current operating status parameters, a real data sequence of any of the power batteries is obtained, sorted by acquisition time. Based on the predicted operating state parameters, a predicted data sequence for any of the power batteries, sorted by acquisition time, is obtained. Calculate the residual sequence by combining the real data sequence and the predicted data sequence; Based on the residual sequence, the root mean square error, residual kurtosis, and residual spectrum dominant frequency energy of the residual characteristics are calculated.

7. A health management device for a power battery, characterized in that, include: The acquisition module is used to acquire the current operating status parameters of any power battery in the vehicle power battery system and determine the current twin model corresponding to any of the power batteries. The deduction module is used to deduce the current twin model in advance to obtain the number of deduction cycles when the capacity of any of the power batteries decays to a preset retirement threshold; The prediction module is used to obtain the predicted operating state parameters output by the current twin model, and to use the predicted operating state parameters and the current operating state parameters to calculate the residual characteristics of any of the power batteries. The control module is used to combine the number of simulation cycles, the residual characteristics, and the current operating status parameters to determine the health status information of any of the power batteries, and to execute battery health management actions that match the health status information.

8. A vehicle, characterized in that, include: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the program to implement the health management method for a power battery as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the health management method for the power battery as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed, it is used to implement the health management method for power batteries as described in any one of claims 1-6.

Images