Method for jointly determining battery health status and remaining life, electronic device, and medium

Abstract

The present invention discloses a method, electronic device, and medium for jointly determining the state of health and remaining life of a battery, and relates to the field of battery management technology. The method includes: obtaining the battery operating parameters at the current moment; constructing a hybrid model based on a physical model and a data-driven model, and controlling the hybrid model through an adaptive gating mechanism to determine the current battery state of health based on the battery operating parameters; the physical model is implemented by introducing a thermal model based on the equivalent circuit model, and the data-driven model is implemented by introducing a lightweight neural network based on the neural network model; the remaining life of the current battery is determined based on the current battery state of health through a life prediction model and an optimization algorithm, the life prediction model is implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction, and the optimization algorithm is implemented based on Bayesian optimization or reinforcement learning. This solution can improve the accuracy of determining the battery state of health and remaining life.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of battery management technology, and in particular to a method for jointly determining the state of health and remaining life of a battery, an electronic device, and a medium. Background Art

[0002] Lithium-ion batteries are widely used in electric vehicles, electric bicycles, and other applications due to their high energy density, long cycle life, and excellent charge-discharge characteristics. Over long-term use, batteries' performance gradually degrades due to factors such as charge-discharge cycles, ambient temperature, and discharge rate, leading to a decrease in health and a shortened remaining lifespan. Therefore, determining the battery's state of health (SOH) and remaining useful life (RUL) is necessary to improve battery safety.

[0003] Currently, battery state of health and remaining life are often determined based on a single model. This can lead to the following issues: single models rely on complex experimental calibration, making them incapable of addressing variations in battery performance under varying operating conditions; single models also place high demands on data quality and lack physical interpretability. These issues can lead to significant errors in the determined battery state of health and remaining life. Summary of the Invention

[0004] The present invention provides a method for jointly determining the state of health and remaining life of a battery, an electronic device, and a medium, which can improve the accuracy of determining the state of health and remaining life of a battery.

[0005] In a first aspect, an embodiment of the present invention provides a method for jointly determining a battery's state of health and remaining life, including:

[0006] Obtaining the current battery operating parameters, wherein the battery operating parameters include at least capacity-related parameters, power and internal resistance-related parameters, differential capacity-related parameters, electrochemical impedance spectroscopy-related parameters, and basic state parameters;

[0007] A hybrid model based on a physical model and a data-driven model is constructed, and the hybrid model is controlled by an adaptive gating mechanism to determine the current battery health state based on the battery operating parameters; the physical model is implemented by introducing a thermal model on the basis of an equivalent circuit model, and the data-driven model is implemented by introducing a lightweight neural network on the basis of a neural network model;

[0008] The remaining life of the current battery is determined based on the current battery health status through a life prediction model and an optimization algorithm. The life prediction model is implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction, and the optimization algorithm is implemented based on Bayesian optimization or reinforcement learning.

[0009] In a second aspect, an embodiment of the present invention provides a device for jointly determining a battery's state of health and remaining life, including:

[0010] an acquisition module, configured to acquire current battery operating parameters, wherein the battery operating parameters include at least capacity-related parameters, power and internal resistance-related parameters, differential capacity-related parameters, electrochemical impedance spectroscopy-related parameters, and basic state parameters;

[0011] A health status determination module is configured to construct a hybrid model based on a physical model and a data-driven model, and to control the hybrid model through an adaptive gating mechanism to determine the current battery health status based on the battery operating parameters; the physical model is implemented by introducing a thermal model based on the equivalent circuit model, and the data-driven model is implemented by introducing a lightweight neural network based on the neural network model;

[0012] A remaining life determination module is used to determine the current remaining life of the battery based on the current battery health state through a life prediction model and an optimization algorithm, wherein the life prediction model is implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction, and the optimization algorithm is implemented based on Bayesian optimization or reinforcement learning.

[0013] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the method according to the first aspect.

[0017] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0018] The technical solution of the embodiment of the present invention, by acquiring battery operating parameters including multi-dimensional features, can provide richer operating condition data support for determining the current battery health state and the current battery remaining life. Through the deep integration of physical models and data-driven models, it combines the physical interpretability of physical models with the high adaptability of data-driven models under different operating conditions, and realizes more accurate determination of the current battery health state in real time. Through degradation curve extrapolation, multi-step rolling prediction, Bayesian optimization, reinforcement learning and other means, a more adaptable and reliable prediction of the current battery remaining life is achieved based on the current battery health state. Therefore, this solution can improve the accuracy of determining the battery health state and remaining life.

[0019] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 This is a flowchart of a method for jointly determining battery health status and remaining life provided in accordance with the first embodiment of the present invention;

[0022] Figure 2 This is a flowchart of a method for jointly determining battery health status and remaining life provided in accordance with a second embodiment of the present invention;

[0023] Figure 3 2 is a schematic diagram of a device for jointly determining a battery's state of health and remaining life according to a third embodiment of the present invention;

[0024] Figure 4 It is a schematic structural diagram of an electronic device implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first," "second," and the like in the present invention are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatuses.

[0027] Example 1

[0028] Figure 1 This is a flowchart of a method for jointly determining the battery state of health and remaining life according to Embodiment 1 of the present invention. This embodiment is applicable to situations where the joint determination of the battery state of health and remaining life is implemented. The method can be performed by an apparatus for jointly determining the battery state of health and remaining life, which can be implemented in software and / or hardware and integrated into an electronic device. Furthermore, electronic devices include, but are not limited to, computers, laptops, servers, and the like.

[0029] like Figure 1 As shown, the method includes:

[0030] S110 , obtaining current battery operating parameters, where the battery operating parameters include at least capacity-related parameters, power and internal resistance-related parameters, differential capacity-related parameters, electrochemical impedance spectroscopy-related parameters, and basic state parameters.

[0031] Battery operating parameters may be parameters related to battery operation. Capacity-related parameters may be battery operating parameters related to battery capacity. Power and internal resistance-related parameters may be battery operating parameters related to the output power capability of the battery. Differential capacity-related parameters may be battery operating parameters related to the differential capacity of the battery. Electrochemical impedance spectroscopy-related parameters may be battery operating parameters related to the electrochemical impedance spectroscopy of the battery. Basic state parameters may be parameters reflecting the basic state of battery operation. It should be noted that the method for obtaining the above parameters is not limited in this step.

[0032] In one embodiment, the capacity-related parameters include the current capacity and the initial capacity; the power and internal resistance-related parameters include the current internal resistance, the initial internal resistance, the current polarization impedance and the polarization time constant; the differential capacity-related parameters include the current peak position and the current peak width in the differential capacity curve; the electrochemical impedance spectroscopy-related parameters include the current interface charge transfer impedance and the current double-layer capacitance; the basic state parameters include the current voltage, the current current and the current temperature.

[0033] The current capacity can be the battery capacity at the current moment, expressed as Q(t). The initial capacity can be the capacity at the time of initial use of the battery, expressed as Q(0). The difference between Q(0) and Q(t) can be understood as the capacity decay. Q(0) and Q(t) can be obtained by measuring the battery capacity at the time of initial use and at the current moment, respectively.

[0034] The current internal resistance can be the internal resistance of the battery at the current moment, expressed as R0(t); the initial internal resistance can be the internal resistance of the battery when it is initially used, expressed as R0(0); R0(0) and R0(t) can be obtained by detecting the internal resistance of the battery when it is initially used and at the current moment, respectively. The current polarization impedance can be the impedance generated by the polarization phenomenon at the current moment, expressed as R p (t) can be obtained through an online impedance measurement unit, which can be hardware that can implement small AC signal injection or pulse discharge. The polarization time constant can be a parameter that describes the speed of the battery polarization process and is expressed as , can be obtained through experimental measurement or theoretical calculation. R0(t), R p (t), It can be used to evaluate the dynamic health status of the battery.

[0035] Differential capacity can be the change in battery capacity corresponding to a unit voltage change during the battery charging and discharging process. Differential capacity curve This is a curve approximated by the ratio of the capacity difference between adjacent data points to the corresponding voltage difference at the current moment. The differential capacity curve can be approximated by sampling a short pulse response curve using the online impedance measurement unit. The peak position (current peak position) and peak width (current peak width) can be extracted from the differential capacity curve.

[0036] Electrochemical Impedance Spectroscopy (EIS) is an electrochemical analysis technique used to study electrochemical systems. It can be used to obtain an approximate electrochemical impedance spectrum by collecting a short pulse response curve using an online impedance measurement unit. The current interface charge transfer impedance, R, can be extracted from the electrochemical impedance spectrum. ct(t), the double-layer capacitance at the current moment, namely the current double-layer capacitance C, can be extracted dl (t), R ct (t) and C dl (t) can be used to characterize the aging characteristics of battery materials.

[0037] The current voltage can be the current battery terminal voltage (V(t)), which can be obtained using a high-precision voltage acquisition sensor. The current current can be the current battery charge / discharge current (I(t)), which can be obtained using a high-precision current acquisition sensor. The current temperature can be the current temperature (T(t), which can include both the ambient temperature and the internal battery temperature. Temperature sensors can be configured both inside and outside the battery to monitor the battery cell temperature and the ambient temperature. Optionally, a humidity sensor or cooling fan speed sensor can be used to obtain more comprehensive thermal management information about the battery.

[0038] During the aforementioned battery operating parameter collection process, the collected data requires preprocessing. This includes outlier detection, which detects possible jumps or drifts in the raw sensor data. Statistical methods (such as isolation forest-based anomaly detection) are then used to eliminate hardware failures or obviously unreasonable data points. Another example is multi-stage noise filtering. Kalman filtering or strong tracking filtering can be used to perform real-time filtering on high-speed sampled V(t) and I(t) to eliminate high-frequency noise. Wavelet decomposition or empirical mode decomposition can be used to process key characteristic signals such as EIS or differential capacity curves to distinguish true aging characteristics from noise interference. Pulse response correction can be performed to ensure the extracted impedance or polarization characteristics are authentic and reliable.

[0039] In this step, the battery operating parameters collected at the current moment can be combined to obtain a feature vector Z(t) to normalize the features of different dimensions and ranges. The expression of Z(t) can be: .

[0040] For the storage of the collected battery operating parameters, a local data storage and cloud synchronization mechanism can be established to ensure both real-time performance and the accumulation of long-term historical data.

[0041] It should be noted that for data collection of battery operating parameters, a digital twin model can be built on the vehicle controller or cloud server to interact with the actual collected data in real time. The digital twin model can record, simulate, and predict the battery's state evolution under different operating conditions and environments, forming a closed loop between data collection and model deduction.

[0042] For data collection of battery operating parameters, different sampling frequencies can be set for different parameters based on actual application needs. For example, in an e-bike riding scenario, a medium frequency (e.g., 1Hz-10Hz) can be set to collect basic state parameters. For data collection of differential capacity curves and electrochemical impedance spectroscopy, a higher frequency or more detailed scan steps can be used within a short period of time to obtain high-resolution impedance or differential capacity data.

[0043] Data collection for battery operating parameters can be achieved through a combination of real-time online testing and periodic offline testing. Real-time online testing continuously collects battery operating parameters during operation. Periodic offline testing involves regularly placing the battery in a laboratory environment for precise capacity testing, EIS measurements, and differential capacity curve recording. This allows for calibration or correction of online impedance measurement units (for example, to compensate for measurement deviations or update reference parameters).

[0044] Optionally, data from multiple scenarios can be collected to determine a benchmark for subsequent offline calibration of battery operating parameters using a hybrid model. In the laboratory, accelerated aging experiments can be designed for different temperatures (e.g., -10°C, 25°C, 40°C), discharge rates (e.g., 0.5C, 1C, 2C), and depths of discharge. Regular, precise capacity testing, internal resistance testing, EIS measurements, and differential capacity curve recording can be performed, serving as a benchmark for subsequent offline calibration of battery operating parameters using a hybrid model.

[0045] Optionally, diverse training and validation sets can be generated by collecting data from multiple operating conditions. For example, during real-world e-bike operation, diverse training and validation sets can be generated by collecting data from various road conditions (e.g., flat roads, uphill slopes, and bumpy roads), usage habits (e.g., rapid acceleration, fully loaded, unloaded), and environmental climates (e.g., hot, cold, and humid). This can then be used to verify the hybrid model's prediction accuracy and robustness under different operating conditions by comparing it with laboratory data or the results of a digital twin model.

[0046] Optionally, when periodically testing the battery offline, the key aging parameters in the model are continuously updated, and the collected large sample data is incorporated into the system database to continuously improve the generalization ability of subsequent model predictions.

[0047] Through comprehensive data acquisition from multiple sources, multiple sensors, and multiple scenarios (combining online and offline, standard sampling with high-resolution pulse / EIS measurements), and employing advanced filtering and feature extraction techniques, we can provide richer and more reliable input data and calibration benchmarks for subsequent determination of the current battery state of health and remaining battery life. This not only enables monitoring of conventional capacity decay, but also captures deeper aging characteristics such as power decay and interfacial impedance changes, significantly improving the accuracy and robustness of health and remaining life predictions.

[0048] S120. Construct a hybrid model based on a physical model and a data-driven model, and control the hybrid model through an adaptive gating mechanism to determine the current battery health status based on the battery operating parameters; the physical model is implemented by introducing a thermal model on the basis of an equivalent circuit model, and the data-driven model is implemented by introducing a lightweight neural network on the basis of a neural network model.

[0049] In the embodiment of the present invention, the physical model can be in the framework of the equivalent circuit model (ECM), except for the internal resistance R0(t) and the polarization impedance R p In addition to (t), a thermal model is introduced to achieve this. The introduction of the thermal model can consider the impact of battery temperature on internal resistance and polarization process, and establish thermoelectric coupling equations (such as energy conservation based on heat capacity and thermal resistance).

[0050] In one embodiment, the physical model is implemented based on the relationship between the current voltage, the open circuit voltage, the current current, the current internal resistance, the current polarization impedance, the polarization time constant, and the additional voltage difference caused by temperature field coupling; wherein, the open circuit voltage is related to the current temperature and a first health state; the first health state is a health state determined by the physical model, which is used to determine the current battery health state.

[0051] Based on the above content, the terminal voltage equation corresponding to the physical model can be formalized as follows:

[0052]

[0053] in, is the additional voltage difference caused by temperature field coupling, which can be determined by the above thermal model; SOHphys(t) is the first health state, that is, the health state determined by the physical model based on the battery operating parameters at the current moment; The remaining parameters in the above terminal voltage equation have been described in S110 and will not be repeated here.

[0054] In embodiments of the present invention, a data-driven model can be implemented by introducing a lightweight neural network based on a neural network model. Specifically, the data-driven model can be a neural network model such as a Long Short-Term Memory network (LSTM) or a Gated Recurrent Unit (GRU). Alternatively, the data-driven model can be a lightweight neural network such as Tiny Machine Learning (TinyML) to reduce computational complexity and improve online adaptability.

[0055] The data-driven model can be a pre-trained model. The feature vector Z(t) obtained by merging the battery operating parameters collected at the current moment is used as the model input. The model can then output the predicted health status of the battery at the current moment, recorded as SOHdata(t).

[0056] In this step, the physical model and the data-driven model can be weighted and fused to construct a hybrid model, which is expressed as follows: .in, is a weight factor, ranging from 0 to 1; SOH(t) is the current battery health status obtained by weighted fusion, that is, the battery health status at the current moment determined by the hybrid model.

[0057] In the hybrid model, due to the high computational complexity of the data-driven model, the weight factor and the call of the data-driven model can be adjusted according to the computing resources and error conditions through an adaptive gating mechanism. For example, if the weight factor is adjusted to 1, the first health state SOHphys(t) determined by the physical model is used as the current battery health state SOH(t). For another example, if the weight factor is adjusted to less than 1, the weighted fusion result of the first health state SOHphys(t) determined by the physical model and SOHdata(t) determined by the data-driven model is used as the current battery health state SOH(t). In this case, the data-driven model can be a neural network model LSTM / GRU or a lightweight neural network TinyML, and can be called according to computing resources.

[0058] S130. Determine the remaining life of the current battery based on the current battery health status through a life prediction model and an optimization algorithm, wherein the life prediction model is implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction, and the optimization algorithm is implemented based on Bayesian optimization or reinforcement learning.

[0059] After completing the real-time estimation of the current battery health status, it is necessary to further predict the current battery remaining life based on the current battery health status. The current battery remaining life can be understood as the remaining life of the battery determined at the current moment. The life prediction model can be a model used to predict the current battery remaining life.

[0060] The life prediction model can be implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction. Degradation curve extrapolation can be understood as extrapolating the current battery health state into the future based on the degradation curve of each parameter in the battery operating parameters collected at the current moment, that is, the curve of each parameter decaying over time, and then determining the moment when the future battery health state drops to the corresponding health state threshold, and taking this moment as the end-of-life moment of the battery, and then taking the difference between the end-of-life moment and the current moment as the current remaining battery life. Multi-step rolling prediction can be understood as using a deep network such as LSTM or GRU to perform a multi-step rolling prediction of the current battery health state, and then determining the end-of-life moment when the battery health state after multi-step rolling drops to the corresponding health state threshold, and then taking the difference between the end-of-life moment and the current moment as the current remaining battery life. The health state threshold can be understood as the health state at which the battery has reached the end of its life or requires maintenance / replacement.

[0061] During the above prediction process, the life prediction model must be optimized using an optimization algorithm. This optimization algorithm can be based on Bayesian Optimization (BO), which improves the model's computational efficiency, or reinforcement learning (RL), which dynamically adjusts model parameters. The specific optimization algorithm used can be determined based on actual computing resources.

[0062] The technical solution of the embodiment of the present invention, by acquiring battery operating parameters including multi-dimensional features, can provide richer operating condition data support for determining the current battery health state and the current battery remaining life. Through the deep integration of physical models and data-driven models, it combines the physical interpretability of physical models with the high adaptability of data-driven models under different operating conditions, and realizes more accurate determination of the current battery health state in real time. Through degradation curve extrapolation, multi-step rolling prediction, Bayesian optimization, reinforcement learning and other means, a more adaptable and reliable prediction of the current battery remaining life is achieved based on the current battery health state. Therefore, this solution can improve the accuracy of determining the battery health state and remaining life.

[0063] Example 2

[0064] Figure 2This is a flowchart of a method for jointly determining the battery health status and remaining life provided in accordance with the second embodiment of the present invention. This embodiment is based on the above-mentioned first embodiment, and further refines the construction of a hybrid model based on a physical model and a data-driven model, and controls the hybrid model through an adaptive gating mechanism to determine the current battery health status based on the battery operating parameters; and further refines the determination of the current battery remaining life based on the current battery health status through a life prediction model and an optimization algorithm.

[0065] like Figure 2 As shown, the method includes:

[0066] S110 , obtaining current battery operating parameters, where the battery operating parameters include at least capacity-related parameters, power and internal resistance-related parameters, differential capacity-related parameters, electrochemical impedance spectroscopy-related parameters, and basic state parameters.

[0067] S121. Construct a hybrid model by weighted fusion of physical model and data-driven model.

[0068] S122. Through an adaptive gating mechanism, when computing resources are limited, the first health state is used as the current battery health state, or the result of weighted fusion of the first health state and the second health state is used as the current battery health state; the first health state is determined by the physical model based on the battery operating parameters, and the second health state is determined by the lightweight neural network based on the battery operating parameters.

[0069] S123. Through an adaptive gating mechanism, when computing resources are not limited, a result of weighted fusion of the first health state and the third health state is used as the current battery health state; the third health state is determined by the neural network model based on the battery operating parameters.

[0070] The following content explains S121-S123 in detail:

[0071] The physical model and the data-driven model are weightedly integrated to construct a hybrid model, which is expressed as follows: .in, is a weight factor with a value of 0 to 1; SOHphys(t) is the first health state determined by the physical model based on the battery operating parameters; SOHdata(t) is the battery health state determined by the data-driven model based on the battery operating parameters, which can be the second health state determined by the lightweight neural network or the third health state determined by the neural network model; SOH(t) is the current battery health state obtained by weighted fusion.

[0072] Through an adaptive gating mechanism, when computing resources are limited, such as in a low-power battery management system (BMS), the physical model is used first and the weighting factor is adjusted to 1. The first health state is used as the current battery health state. When the error in the physical model result is large, the physical model and the lightweight neural network are fused, and the weighting factor is adjusted to less than 1. The result of the weighted fusion of the first health state and the second health state is used as the current battery health state.

[0073] In one embodiment, when computing resources are limited, if the error of the first health state is lower than a set error threshold, the first health state is used as the current battery health state; otherwise, the result of weighted fusion of the first health state and the second health state is used as the current battery health state. The error of the first health state can be the error obtained by comparing the first health state with the actual battery health state; the set error threshold is not limited and can be set according to actual application needs.

[0074] It should be noted that the lightweight neural network TinyML can run efficiently in embedded devices (such as BMS). The optimization strategies for lightweight neural networks include: model quantization, which converts 32-bit floating-point numbers into 8-bit integers to improve inference speed. The quantized model can run on low-power devices and adapt to resource-constrained scenarios; model pruning, which reduces computational complexity and improves online inference performance by removing redundant connections in the neural network. The pruning ratio can be dynamically adjusted to adapt to the computing power requirements of different embedded hardware.

[0075] Through the adaptive gating mechanism, when computing resources are not limited, such as cloud computing, the physical model and the neural network model can be integrated, and the weight factor can be adjusted to be less than 1, that is, the result of the weighted fusion of the first health state and the third health state can be used as the current battery health state.

[0076] Optionally, during the process of determining the third health state using the neural network model or the second health state using the lightweight neural network, incremental learning can be employed to improve adaptability to different operating conditions. This involves updating the neural network weights in small, real-time steps, updating only key layers of the model (such as the last layer of an LSTM / GRU) rather than retraining the entire network, thus reducing computational requirements. Based on the pre-trained model, some weights can be fine-tuned under new operating conditions to improve adaptability and avoid overfitting. Normalization and data dimensionality reduction techniques can also be employed to remove redundant information and improve computational efficiency.

[0077] Optionally, adaptive unscented Kalman filtering and particle filtering can be used to dynamically correct the current battery health state. The adaptive unscented Kalman filter uses the current battery health state as a state variable and uses observed data for filtering correction to reduce the impact of noise. It also improves filtering accuracy by adaptively adjusting the filter noise covariance. The particle filter can estimate the probability of the current battery health state in nonlinear, non-Gaussian noise environments, improving prediction stability. When the current battery health state estimate shows significant deviations, it can trigger particle resampling to improve calculation accuracy.

[0078] In one embodiment, the second health state or the third health state includes the following types: a capacity-type health state determined based on the capacity-related parameters; a power-type health state determined based on the power and internal resistance-related parameters; a differential capacity-type health state determined based on the differential capacity-related parameters; an electrochemical impedance spectroscopy-type health state determined based on the electrochemical impedance spectroscopy-related parameters; and a multidimensional health state determined based on joint probability modeling of the capacity-type health state and the power-type health state.

[0079] Among them, capacity-based health status SOH cap (t) can be expressed as: The meaning of the parameters has been shown in Example 1 and will not be repeated here.

[0080] Power-based State of Health (SOH) power (t) can be expressed as: The meaning of the parameters has been shown in Example 1 and will not be repeated here.

[0081] Differential Capacity State of Health (SOH) dQ / dV (t) can be expressed as follows, where is the characteristic peak offset determined based on the current peak position, is the increase in half-peak width based on the current peak width, .

[0082] Electrochemical impedance spectroscopy-based state of health SOH EIS (t) can be expressed as follows, where the meanings of the parameters are shown in Example 1. .

[0083] Based on the capacity-based health status and power-based health status, a Bayesian fusion method can be used to construct a joint probability distribution as follows. Through joint probability modeling, a consistent estimate of health status in different dimensions can be obtained to obtain a multi-dimensional health status. Here, Z1:t can be understood as the feature vector obtained by combining the battery operating parameter features collected from time 1 (not limited) to the current time t.

[0084]

[0085] S131 , predicting the end-of-life moment when the current battery health state drops to a corresponding health state threshold multiple times using a life prediction model, and optimizing the life prediction model in combination with an optimization algorithm during the prediction process.

[0086] In one embodiment, a life prediction model implemented based on degradation curve extrapolation is used to extrapolate the battery operating parameters into the future, and in combination with the hybrid model, determine the end-of-life moment based on the extrapolated battery operating parameters; a life prediction model implemented based on multi-step rolling prediction is used to perform multi-step rolling prediction on the current battery health status using a neural network model or a lightweight neural network to determine the end-of-life moment; an optimization algorithm implemented based on Bayesian optimization is used for optimization when computing resources are not limited; an optimization algorithm implemented based on reinforcement learning is used for optimization when computing resources are limited.

[0087] S132: Determine the current remaining battery life based on the difference between the life expiration time and the current time, wherein the current remaining battery life is expressed in the form of a confidence interval.

[0088] The following content explains S131-S132 in detail:

[0089] If the degradation curves of the current battery operating parameters, such as the current capacity Q(t), current internal resistance R0(t), and EIS parameters, over time / cycle number, can be well fitted using a fixed function form (such as exponential decay, logarithmic decay, or a double exponential model), the current battery operating parameters can be directly extrapolated to the future. For example, the extrapolated Q(t) and R0(t) can be expressed as follows, where: and represents the fitted decay function.

[0090]

[0091] After obtaining the extrapolated battery operating parameters, the current battery health status can be determined through the hybrid model in the above steps, and the future battery health status can be determined based on the extrapolated battery operating parameters. Then, the moment when the future battery health status falls to the corresponding health status threshold can be determined, and this moment can be used as the end of battery life.

[0092] The life prediction model based on multi-step rolling prediction can use a neural network model or a lightweight neural network to perform a multi-step rolling prediction on the current battery health state, and then determine the battery health state after the multi-step rolling, and determine the end-of-life time when the battery health state after the multi-step rolling falls to the corresponding health state threshold. Taking the neural network model as LSTM as an example, the multi-step rolling prediction can be expressed as follows, where k is the number of prediction steps, are the feature vectors of the prediction input model for each step from t to t+k, The battery health status after multiple scrolling steps.

[0093]

[0094] In practical applications, the degradation curve extrapolation can be used in conjunction with the multi-step rolling prediction to take into account both the stability of the extrapolation and the sensitivity of the neural network to abnormal conditions.

[0095] It should be noted that different types of battery health status may correspond to different health status thresholds. For example, the health status threshold corresponding to the capacity-type health status may be a capacity threshold (such as 80% or 70%), which is considered to indicate that the capacity has reached an unacceptable level of attenuation. This threshold can be flexibly set according to the vehicle's range or actual needs. Another example is that the health status threshold corresponding to the power-type health status may be a power threshold (such as 70%), which is considered to indicate that the power performance cannot meet the usage requirements. In actual applications, one or more thresholds can be set at the same time. When the health status of any type of battery first falls below the corresponding threshold, it can be considered that the battery has reached the end of its life or requires maintenance / replacement.

[0096] It should be noted that no matter which health status threshold is used to determine the end-of-life time of the battery, the difference between the end-of-life time and the current time can be used as the current remaining battery life.

[0097] It should be noted that during the prediction process of the above-mentioned life prediction model, Monte Carlo sampling or posterior sampling based on particle filtering can be introduced to make multiple predictions of the current battery's remaining life, and ultimately output the current battery's remaining life in the form of a confidence interval. For example, when the current battery's remaining life is output, interval information such as the most likely value and upper and lower confidence boundaries (such as a 95% confidence interval) can be provided to facilitate the provision of safety margins when making operation and maintenance decisions.

[0098] To improve the computational efficiency and online adaptability of current battery remaining life prediction, an optimization algorithm was introduced to optimize the life prediction model. An adaptive optimization strategy was employed, selecting different optimization algorithms based on different computing environments. When computing resources were not limited, the LSTM / GRU neural network models were used for RUL prediction, combined with Bayesian optimization for global optimization. When computing resources were limited, the lightweight neural network TinyML was used for RUL prediction, combined with reinforcement learning for local optimization to reduce computational overhead.

[0099] Among them, Bayesian optimization does not require a large amount of sampling and only requires 10 to 50 iterations to obtain an approximate optimal solution. It uses Gaussian process regression to model the parameter distribution, improves optimization efficiency, and is suitable for real-time optimization.

[0100] Among them, reinforcement learning can dynamically adjust optimization parameters to improve the adaptability of the model in different environments. Specifically: using reinforcement learning models such as Deep Q-Network (DQN) to learn the optimal optimization parameters under different working conditions; dynamically adjusting the learning rate, step size, attenuation parameters, etc. in the RUL prediction process to optimize online prediction efficiency; compared with traditional optimization algorithms, reinforcement learning can quickly find the optimal solution for optimization parameters with limited computing resources.

[0101] The process of reinforcement learning can be achieved through the following function express:

[0102]

[0103] in, Represents the model parameters to be optimized (such as degradation curve parameters, neural network weights, etc.); SOHpred and SOHtrue are the predicted value and actual value of SOH respectively; RULpred and RULtrue are the predicted value and actual value of RUL respectively; and They represent the optimization weights of the prediction error of SOH and the optimization weights of the prediction error of RUL respectively.

[0104] In practical applications, a reward mechanism can be adopted. A decrease in the battery life can provide positive incentives to improve the optimization effect. Through reinforcement learning training, the model parameters for predictions in different battery usage scenarios can be optimized.

[0105] In practical applications, a closed-loop feedback mechanism can be used to continuously optimize RUL predictions. Specifically, if there is a significant deviation between the predicted RUL and the actual value, the following triggers: physical model updates, such as adjusting the SOH degradation parameters to better match the actual degradation curve; data-driven model updates, using incremental learning to update the model parameters of the data-driven model; and optimization algorithm adjustments, such as reinforcement learning to adjust hyperparameters and improve model adaptability.

[0106] In practical applications, the predicted RUL can be used to optimize energy management strategies. If the RUL drops too quickly, the maximum discharge rate can be adjusted to extend battery life. In applications such as shared electric bicycles and battery rentals, battery scheduling strategies can be adjusted based on the predicted RUL to optimize usage costs.

[0107] Optionally, the prediction results can be visualized and evaluated. Specifically, the predicted RUL dynamic change curve over time is compared with the actual aging observation value to evaluate the prediction error and confidence interval. A visual dashboard is provided to display the current RUL prediction value, upper and lower bounds, and key health indicators for vehicle management or user queries.

[0108] Optionally, an online update strategy can be configured. Specifically, based on the latest sensor data and regular offline calibration results, model parameters and algorithm hyperparameters can be continuously iterated to ensure continuous prediction accuracy throughout the lifecycle. If a sudden change in operating conditions (such as extreme temperature or unexpected failure) is detected, reinforcement learning or emergency correction mechanisms can be triggered within a short period of time to perform an emergency update to the RUL and alert the user or management system.

[0109] The technical solution of the embodiment of the present invention controls the weights of the physical model and the data-driven model in the hybrid model and the specific model used by the data-driven model in different computing resource scenarios through an adaptive gating mechanism. This can dynamically balance computing accuracy and computing power consumption, improving the feasibility of practical applications. By combining multi-level degradation curve extrapolation, deep learning multi-step prediction and optimization algorithms, combined with interval prediction and multi-objective adaptive adjustment, it can achieve accurate, dynamic, and interpretable estimation of the remaining battery life. This solution can achieve high-precision, low-computational cost, and highly adaptable joint prediction of SOH and RUL.

[0110] The technical solutions of the embodiments of the present invention can be applied to multiple application scenarios such as intelligent battery management systems, shared electric vehicles, battery rentals, energy storage systems, etc., and can effectively improve the accuracy, real-time performance and safety of battery life prediction, which is of great significance for improving battery life, optimizing operating costs and improving user experience.

[0111] The technical solution of the embodiment of the present invention can have the following advantages:

[0112] 1. Adaptive computing load distribution improves adaptability to diverse computing environments. This invention dynamically adjusts computing methods for cloud-based, high-computing embedded devices, and low-power BMS devices. Through intelligent load switching, it prioritizes the use of physical models when resources are limited. In high-computing environments, it uses LSTM / GRU and combines Bayesian optimization (BO) for global parameter optimization, improving computing efficiency and prediction accuracy.

[0113] 2. Reinforcement learning (RL) optimizes SOH estimation and improves predictive adaptability. This invention utilizes reinforcement learning models, such as deep Q networks, to dynamically adjust SOH estimation parameters, improving prediction accuracy under different operating conditions. Reinforcement learning can intelligently optimize calculation step size and model hyperparameters, reducing computational complexity and enhancing the system's real-time performance and adaptability.

[0114] 3. Multi-scale RUL prediction, supporting both short-term and long-term lifespan predictions. Short-term predictions can use unscented Kalman filtering to optimize load fluctuations. Long-term predictions can combine LSTM + BO and provide confidence intervals through Monte Carlo simulation, improving prediction stability and accuracy.

[0115] 4. SOH mutation warning based on anomaly detection improves safety. This invention can introduce an anomaly detection algorithm to monitor SOH mutations in real time and dynamically adjust the prediction model based on an adaptive threshold strategy to prevent extreme aging or thermal runaway. At the same time, it can send an alarm to the BMS to optimize the charge and discharge strategy and improve battery safety.

[0116] 5. Full lifecycle battery management and optimized battery operation and maintenance. This invention combines SOH and RUL predictions to dynamically optimize charge and discharge strategies, reducing the impact of over-discharge on battery life. In scenarios such as shared electric vehicles and battery rentals, intelligent battery replacement and scheduling based on RUL predictions improve operational efficiency. Historical data is combined with retrospective analysis of battery aging trends to optimize long-term maintenance strategies.

[0117] 6. Incorporating external environmental variables to improve prediction generalization capabilities. The present invention introduces external variables such as temperature to improve prediction accuracy.

[0118] Example 3

[0119] Figure 3 This is a schematic diagram of a device for jointly determining the state of health and remaining life of a battery according to the third embodiment of the present invention. This embodiment is applicable to the case where the state of health and remaining life of a battery are jointly determined. Figure 3 As shown, the specific structure of the device includes:

[0120] An acquisition module 31 is configured to acquire current battery operating parameters, wherein the battery operating parameters include at least capacity-related parameters, power and internal resistance-related parameters, differential capacity-related parameters, electrochemical impedance spectroscopy-related parameters, and basic state parameters.

[0121] A health status determination module 32 is configured to construct a hybrid model based on a physical model and a data-driven model, and to control the hybrid model through an adaptive gating mechanism to determine the current battery health status based on the battery operating parameters; the physical model is implemented by introducing a thermal model on the basis of an equivalent circuit model, and the data-driven model is implemented by introducing a lightweight neural network on the basis of a neural network model;

[0122] The remaining life determination module 33 is used to determine the current remaining life of the battery based on the current battery health state through a life prediction model and an optimization algorithm, wherein the life prediction model is implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction, and the optimization algorithm is implemented based on Bayesian optimization or reinforcement learning.

[0123] This embodiment provides a device for jointly determining the battery state of health and remaining life. An acquisition module acquires the current battery operating parameters, which include at least capacity-related parameters, power-and-internal-resistance-related parameters, differential capacity-related parameters, electrochemical impedance spectroscopy-related parameters, and basic state parameters. A health state determination module constructs a hybrid model based on a physical model and a data-driven model, and controls the hybrid model through an adaptive gating mechanism to determine the current battery state of health based on the battery operating parameters. The physical model is implemented by introducing a thermal model based on an equivalent circuit model, and the data-driven model is implemented by introducing a lightweight neural network based on a neural network model. A remaining life determination module determines the current battery remaining life based on the current battery state of health using a life prediction model and an optimization algorithm. The life prediction model is implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction, and the optimization algorithm is implemented based on Bayesian optimization or reinforcement learning. This solution can improve the accuracy of determining the battery state of health and remaining life.

[0124] Furthermore, the capacity-related parameters include current capacity and initial capacity;

[0125] The power and internal resistance related parameters include current internal resistance, initial internal resistance, current polarization impedance and polarization time constant;

[0126] The differential capacity related parameters include the current peak position and the current peak width in the differential capacity curve;

[0127] The electrochemical impedance spectroscopy related parameters include the current interface charge transfer impedance and the current double layer capacitance;

[0128] The basic state parameters include current voltage, current current and current temperature.

[0129] Furthermore, the physical model is implemented based on the relationship between the current voltage, the open circuit voltage, the current current, the current internal resistance, the current polarization impedance, the polarization time constant, and the additional voltage difference caused by temperature field coupling;

[0130] The open circuit voltage is related to the current temperature and a first health state; the first health state is a health state determined by the physical model and is used to determine the current battery health state.

[0131] Furthermore, the health status determination module 32 is specifically configured to:

[0132] Constructing a hybrid model through weighted fusion of physical model and data-driven model;

[0133] Through an adaptive gating mechanism, when computing resources are limited, a first health state is used as the current battery health state, or a weighted fusion result of the first health state and the second health state is used as the current battery health state; the first health state is determined by the physical model based on the battery operating parameters, and the second health state is determined by the lightweight neural network based on the battery operating parameters;

[0134] Through an adaptive gating mechanism, when computing resources are not limited, a weighted fusion result of the first health state and the third health state is used as the current battery health state; the third health state is determined by the neural network model based on the battery operating parameters.

[0135] Furthermore, when computing resources are limited, if the error of the first health state is lower than the set error threshold, the first health state is used as the current battery health state; otherwise, the result of the weighted fusion of the first health state and the second health state is used as the current battery health state.

[0136] Furthermore, the second health status or the third health status includes the following types:

[0137] A capacity-type health state determined based on the capacity-related parameters; a power-type health state determined based on the power and internal resistance-related parameters; a differential capacity-type health state determined based on the differential capacity-related parameters; an electrochemical impedance spectroscopy-type health state determined based on the electrochemical impedance spectroscopy-related parameters; and a multidimensional health state determined based on the joint probability modeling of the capacity-type health state and the power-type health state.

[0138] Furthermore, the remaining life determination module 33 is specifically configured to:

[0139] Predicting the end-of-life moment when the current battery health state falls to a corresponding health state threshold multiple times using a life prediction model, and optimizing the life prediction model in combination with an optimization algorithm during the prediction process;

[0140] The current remaining battery life is determined based on the difference between the end-of-life moment and the current moment, and the current remaining battery life is expressed in the form of a confidence interval.

[0141] Furthermore, a life prediction model based on degradation curve extrapolation is used to extrapolate the battery operating parameters into the future, and to determine the end of life time based on the extrapolated battery operating parameters in combination with the hybrid model;

[0142] A life prediction model based on multi-step rolling prediction is used to perform a multi-step rolling prediction on the current battery health state using a neural network model or a lightweight neural network to determine the end of life time;

[0143] An optimization algorithm based on Bayesian optimization, used for optimization when computing resources are not limited;

[0144] An optimization algorithm based on reinforcement learning, used for optimization under limited computing resources.

[0145] The device for jointly determining the battery health status and remaining life provided in an embodiment of the present invention can execute the method for jointly determining the battery health status and remaining life provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0146] Example 4

[0147] Figure 4 is a schematic diagram of the structure of an electronic device that implements an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0148] like Figure 4As shown, electronic device 10 includes at least one processor 11 and memory, such as read-only memory (ROM) 12 and random access memory (RAM) 13, communicatively connected to at least one processor 11. The memory stores computer programs executable by the at least one processor. Processor 11 can perform various appropriate actions and processes based on the computer programs stored in ROM 12 or loaded from storage unit 18 into RAM 13. RAM 13 can also store various programs and data required for the operation of electronic device 10. Processor 11, ROM 12, and RAM 13 are interconnected via bus 14. An input / output (I / O) interface 15 is also connected to bus 14.

[0149] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0150] The processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any other suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the method for jointly determining the battery health status and remaining life.

[0151] In some embodiments, the method for jointly determining the state of health and remaining life of a battery may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for jointly determining the state of health and remaining life of a battery may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the method for jointly determining the state of health and remaining life of a battery via any other suitable means (e.g., via firmware).

[0152] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0153] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0154] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or apparatus. A computer-readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or apparatus, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0155] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device that has: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0156] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0157] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0158] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0159] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for jointly determining battery health status and remaining life, characterized in that: include: Obtaining the current battery operating parameters, wherein the battery operating parameters include at least capacity-related parameters, power and internal resistance-related parameters, differential capacity-related parameters, electrochemical impedance spectroscopy-related parameters, and basic state parameters; A hybrid model based on a physical model and a data-driven model is constructed, and the hybrid model is controlled by an adaptive gating mechanism to determine the current battery health state based on the battery operating parameters; the physical model is implemented by introducing a thermal model on the basis of an equivalent circuit model, and the data-driven model is implemented by introducing a lightweight neural network on the basis of a neural network model; Determining the current remaining battery life based on the current battery health state using a life prediction model and an optimization algorithm, wherein the life prediction model is implemented based on at least one of degradation curve extrapolation and multi-step rolling prediction, and the optimization algorithm is implemented based on Bayesian optimization or reinforcement learning; The method includes: constructing a hybrid model based on a physical model and a data-driven model, and controlling the hybrid model to determine the current battery health state based on the battery operating parameters through an adaptive gating mechanism, including: Constructing a hybrid model through weighted fusion of physical model and data-driven model; Through an adaptive gating mechanism, when computing resources are limited, a first health state is used as the current battery health state, or a weighted fusion result of the first health state and the second health state is used as the current battery health state; the first health state is determined by the physical model based on the battery operating parameters, and the second health state is determined by the lightweight neural network based on the battery operating parameters; Through an adaptive gating mechanism, when computing resources are not limited, a weighted fusion result of the first health state and the third health state is used as the current battery health state; the third health state is determined by the neural network model based on the battery operating parameters.

2. The method according to claim 1, characterized in that The capacity-related parameters include current capacity and initial capacity; The power and internal resistance related parameters include current internal resistance, initial internal resistance, current polarization impedance and polarization time constant; The differential capacity related parameters include the current peak position and the current peak width in the differential capacity curve; The electrochemical impedance spectroscopy related parameters include the current interface charge transfer impedance and the current double layer capacitance; The basic state parameters include current voltage, current current and current temperature.

3. The method according to claim 2, characterized in that The physical model is implemented based on the relationship between the current voltage, the open circuit voltage, the current current, the current internal resistance, the current polarization impedance, the polarization time constant, and the additional voltage difference caused by temperature field coupling; The open circuit voltage is related to the current temperature and a first health state; the first health state is a health state determined by the physical model and is used to determine the current battery health state.

4. The method according to claim 1, wherein When computing resources are limited, if the error of the first health state is lower than the set error threshold, the first health state is used as the current battery health state; otherwise, the result of the weighted fusion of the first health state and the second health state is used as the current battery health state.

5. The method according to claim 1, wherein The second health state or the third health state includes the following types: Capacity-type health status determined based on the capacity-related parameters; power-type health status determined based on the power and internal resistance-related parameters; differential capacity-type health status determined based on the differential capacity-related parameters; electrochemical impedance spectroscopy-type health status determined based on the electrochemical impedance spectroscopy-related parameters; The multi-dimensional health state is determined based on joint probability modeling of the capacity-based health state and the power-based health state.

6. The method according to claim 1, characterized in that Determining the remaining battery life based on the current battery health state using a life prediction model and an optimization algorithm includes: Predicting the end-of-life moment when the current battery health state falls to a corresponding health state threshold multiple times using a life prediction model, and optimizing the life prediction model in combination with an optimization algorithm during the prediction process; The current remaining battery life is determined based on the difference between the end-of-life moment and the current moment, and the current remaining battery life is expressed in the form of a confidence interval.

7. The method according to claim 6, characterized in that A life prediction model based on degradation curve extrapolation, used to extrapolate the battery operating parameters into the future, and determine the end of life time based on the extrapolated battery operating parameters in combination with the hybrid model; A life prediction model based on multi-step rolling prediction is used to perform a multi-step rolling prediction on the current battery health state using a neural network model or a lightweight neural network to determine the end of life time; An optimization algorithm based on Bayesian optimization, used for optimization when computing resources are not limited; An optimization algorithm based on reinforcement learning, used for optimization under limited computing resources.

8. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images