Power battery recession mechanism analysis and state-of-health estimation method

Abstract

The invention discloses a power battery recession mechanism analysis and state-of-health estimation method, and particularly relates to the technical field of battery management, and the method comprises the steps: S1, synchronously collecting ultrasonic scanning, distributed temperature sensing and electrochemical impedance spectroscopy signals of a battery, S2, extracting and fusing multi-physical field characteristics, constructing an acoustic-thermal-electric coupling characteristic spectrum, and S3, calculating the power battery recession mechanism according to the acoustic-thermal-electric coupling characteristic spectrum. S3, decoupling and quantifying an internal decline mechanism based on the characteristic spectrum, and S4, cooperatively estimating the state of health of the battery through an adaptive fusion algorithm in combination with a mechanism model and a data driving model. According to the method, online, lossless and quantitative analysis of key decline mechanisms such as lithium precipitation and cracks is realized, the health state estimation precision and the early warning capability are remarkably improved, and a complete solution is provided for accurate management of the battery.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and more specifically, to a method for analyzing the degradation mechanism and estimating the health status of a power battery. Background Technology

[0002] With the rapid development of electric vehicles and large-scale energy storage industries, the long-term operational reliability and safety of power batteries have become core concerns. During the cycle of use, batteries undergo complex physicochemical changes, leading to irreversible performance degradation. Therefore, accurately assessing the health status of batteries is key to achieving efficient battery management, while in-depth analysis of their internal degradation mechanisms is the fundamental basis for achieving accurate assessment, early warning, and lifespan prediction.

[0003] Currently, widely studied battery health state estimation methods can be mainly categorized into three types: data-driven methods, model-driven methods, and feature correlation methods. These methods generally rely on macroscopic measurable signals outside the battery (such as voltage, current, and surface temperature) and infer the internal state by indirectly processing or fitting these signals.

[0004] However, their fundamental limitation lies in the lack of an effective means to directly perceive the internal multi-physics field state online, non-destructively, and with spatial resolution under actual battery operating conditions. This makes it difficult for existing technologies to distinguish and quantitatively analyze various coupled degradation mechanisms such as lithium deposition, loss of electrode active materials, particle crack propagation, and SEI film growth. Consequently, the health status estimation results suffer from insufficient early warning capabilities, poor physical interpretability, and weak extrapolation under varying operating conditions.

[0005] In view of this, in order to overcome the above-mentioned shortcomings in the prior art, the present invention provides a method for analyzing the degradation mechanism and estimating the health status of power batteries. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method for analyzing the degradation mechanism and estimating the health status of power batteries, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for analyzing the degradation mechanism and estimating the health status of a power battery, specifically including the following steps: S1. During battery operation, multi-physics field in-situ sensing signals are collected synchronously, including at least actively excited ultrasonic scanning signals, distributed temperature sensing signals, and electrochemical impedance spectroscopy signals. S2. Characteristic parameters that can characterize the internal mechanical structure, thermal behavior and electrochemical reaction kinetics of the battery are extracted from ultrasonic scanning signals, distributed temperature sensing signals and electrochemical impedance spectroscopy signals respectively, and then fused to construct a spatiotemporally correlated acoustic-thermal-electric coupling characteristic spectrum. S3. Based on the acoustic-thermal-electric coupling characteristic spectrum, the degradation mechanism inside the battery is decoupled and quantitatively analyzed to obtain a quantitative index of at least one degradation mechanism. S4. Using quantitative indicators of degradation mechanisms, combined with mechanism models and data-driven models, the health status of the battery is estimated collaboratively through information fusion algorithms.

[0008] Preferably, in step S1, the synchronous acquisition specifically involves: triggering a coordinated measurement cycle during the battery's resting phase or the voltage plateau phase of constant current charging, and controlling the ultrasonic excitation unit, impedance excitation unit, and temperature acquisition unit to perform millisecond-level time-synchronized signal acquisition.

[0009] Preferably, in step S2, extracting feature parameters from the ultrasonic scanning signal includes: processing the received ultrasonic echo signal and extracting acoustic feature vectors related to changes in the internal microstructure of the battery, wherein the acoustic feature vectors include at least one or more of the following: sound wave propagation speed through each layer of the battery, signal energy attenuation coefficient, spectral amplitude of a specific frequency band, or scattering intensity.

[0010] Preferably, in step S2, extracting feature parameters from the distributed temperature sensing signal includes: calculating a thermal feature vector based on temperature sensor readings arranged at multiple locations on the battery surface, wherein the thermal feature vector includes at least one or more of the following: temperature field spatial gradient, maximum temperature rise rate, and temperature distribution non-uniformity index.

[0011] Preferably, in step S2, extracting characteristic parameters from the electrochemical impedance spectroscopy signal includes: analyzing the measured impedance spectrum and extracting electrochemical feature vectors, wherein the electrochemical feature vectors include at least one or more of the following: ohmic internal resistance, charge transfer resistance, Warburg diffusion coefficient, and phase angle at one or more characteristic frequency points.

[0012] Preferably, in step S2, the fusion and construction of the acoustic-thermal-electric coupling feature spectrum specifically involves splicing and aligning the acoustic feature vectors, thermal feature vectors, and electrochemical feature vectors that are spatially related at the same time stamp to form a multidimensional feature matrix, which serves as the coupling feature spectrum characterizing the overall state of the battery at that moment.

[0013] Preferably, step S3 specifically includes: S3.1. Establish a standard characteristic spectrum template library corresponding to different dominant decay mechanisms in advance through experiments; S3.2. Match and decompose the acoustic-thermal-electric coupling characteristic spectrum constructed online with the standard characteristic spectrum template library, and calculate the contribution coefficient of each decay mechanism template to the current battery state. The contribution coefficient is the quantitative index of the decay mechanism.

[0014] Preferably, the dominant degradation mechanism includes at least two of the following: solid electrolyte interfacial film growth, lithium metal deposition, loss of active material, and particle crack propagation in electrode material. The quantitative indicators include the lithium deposition index for characterizing the severity of lithium deposition and / or the crack index for characterizing the propagation of structural cracks.

[0015] Preferably, step S4 specifically includes: S4.1 Establish a battery electrochemical-mechanical coupled degradation model that includes quantitative indicators of degradation mechanism as state variables, update the model using quantitative indicators, and obtain the first health state estimate and the first uncertainty based on the mechanism model. S4.2 Using the acoustic-thermal-electric coupling characteristic spectrum or its principal components after dimensionality reduction as input, the second health state estimate and the second uncertainty are obtained through a data-driven model; S4.3 Based on the first uncertainty and the second uncertainty, an adaptive weighted fusion algorithm is used to fuse the first health state estimate and the second health state estimate, and output the final battery health state estimate and confidence interval.

[0016] Preferably, in step S4.3, the adaptive weighted fusion algorithm is a fusion framework based on Kalman filtering or Bayesian estimation, and its weights are dynamically adjusted according to the estimation error covariance of each model under the current operating conditions.

[0017] The technical effects and advantages of this invention are as follows: 1. This invention provides a direct and information-rich observation window for online and non-destructive observation of the internal state of a battery by simultaneously acquiring ultrasonic, temperature field and electrochemical impedance signals and fusing them to construct an acoustic-thermal-electric coupling characteristic spectrum, overcoming the indirectness and limitations of traditional methods that rely solely on external macroscopic signals. 2. Based on coupled characteristic spectra and a pre-established mechanism template library, this invention can decouple and quantify key degradation mechanisms such as lithium plating and crack propagation, providing an unprecedentedly refined understanding of battery health management and realizing the visualization and quantification of the degradation process; 3. This invention introduces quantitatively analyzed mechanistic indicators as strong physical constraints into the mechanistic model, while using high-dimensional feature spectra to drive the data model. Through an adaptive fusion algorithm based on uncertainty, the estimation results of the two are dynamically balanced, giving full play to the respective advantages of the mechanistic model's strong interpretability and good extrapolation and the data-driven model's high fitting accuracy. This framework enables the health status estimation results to have high accuracy, strong robustness and early warning capability. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the overall steps of the present invention.

[0019] Figure 2 This is a flowchart of the quantitative analysis of the decay mechanism in step S3 of the present invention.

[0020] Figure 3 This is a flowchart of the health status collaborative estimation in step S4 of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0022] This invention provides a method for analyzing the degradation mechanism and estimating the health status of power batteries. The core of this method is to construct a high-dimensional coupled feature spectrum by sequentially collecting and fusing acoustic, thermal, and electrical multi-physics field signals, thereby achieving quantitative analysis of the internal degradation mechanism of the battery and, based on this, achieving high-precision and highly interpretable health status estimation. The hardware system required to implement this method mainly includes the following modules: The main control and synchronization unit, as the core of the system, is responsible for controlling the timing of the entire measurement process, ensuring millisecond-level synchronization of multi-channel signal acquisition, and performing preliminary data processing and execution of core algorithms. This unit has a built-in high-precision clock, which serves as the time reference for all signal acquisitions. An ultrasonic excitation and receiving array, consisting of multiple miniature piezoelectric ultrasonic transducers, preferably in the frequency range of 0.5-10MHz, is integrated on the surface of a single battery cell. The main control and synchronization unit controls the transmission of pulsed ultrasonic waves at a specific frequency and receives echo signals that penetrate or are reflected from the internal interface. A distributed temperature sensor network consists of multiple high-precision thin-film temperature sensors, which are closely attached to the surface of the battery casing in an array, especially at key locations such as the positive and negative tabs, the center of the cell, and the four corners, to obtain the two-dimensional temperature field distribution on the battery surface. The electrochemical impedance spectroscopy measurement unit can inject a set of multi-frequency small-amplitude AC excitation current signals into the battery during battery charging and discharging intervals or under low current conditions, and simultaneously measure its voltage response, thereby calculating the battery's impedance spectrum within a specific frequency range. The main control and synchronization unit coordinates the work of the three sensing modules to ensure strict time alignment of data, laying the foundation for subsequent feature fusion.

[0023] As attached Figure 1-3 As shown, the specific implementation method is as follows: S1. During battery operation, multi-physics field in-situ sensing signals are collected synchronously, including at least actively excited ultrasonic scanning signals, distributed temperature sensing signals, and electrochemical impedance spectroscopy signals. In the specific implementation of this step, synchronous acquisition is crucial for achieving accurate signal fusion. This step is preferably performed when the battery is in a static state (the voltage has stabilized after charging and discharging) or at the voltage plateau stage of constant current charging. At this time, the internal polarization state of the battery is relatively stable and the interference is small. In actual operation, the main control and synchronization unit trigger a coordinated measurement cycle: a command is sent to the ultrasonic excitation unit to emit a predefined ultrasonic pulse, and almost simultaneously, a command is sent to the impedance measurement unit to start injecting impedance spectrum measurement signals. Throughout the measurement cycle, data from the distributed temperature sensor network is acquired at high speed. All signal acquisition is based on the high-precision clock of the main control and synchronization unit to achieve millisecond-level time synchronization, ensuring that the acquired data reflects the battery state at the same instant.

[0024] S2. Characteristic parameters that can characterize the internal mechanical structure, thermal behavior and electrochemical reaction kinetics of the battery are extracted from ultrasonic scanning signals, distributed temperature sensing signals and electrochemical impedance spectroscopy signals respectively, and then fused to construct a spatiotemporally correlated acoustic-thermal-electric coupling characteristic spectrum. This step is crucial for information extraction and fusion, and in its specific implementation: S2.1 Acoustic Feature Extraction: Processing the ultrasonic echo signal; By measuring the time difference of flight of ultrasound through layers of known thickness in the battery (such as from the casing to the negative electrode current collector). Combined with known distance Calculate the sound velocity of sound waves propagating in this layer of material. Changes in sound velocity can sensitively reflect changes in material density and modulus, and are related to SEI thickening and changes in electrode porosity; Compare the amplitude of the transmitted signal With the amplitude energy of the received signal Calculate the signal energy attenuation coefficient (Unit: dB / mm). Increased attenuation may indicate an increase in internal microscopic scatterers (such as lithium dendrites, grain cracks); Perform Fast Fourier Transform or Wavelet Transform on the echo signal to extract the spectral amplitude or scattering intensity of a specific frequency band (such as the high-frequency band). The attenuation or enhancement of high-frequency components is closely related to the scale distribution of the microstructure; The multiple acoustic parameters calculated above are arranged according to the measurement path or spatial location to form an acoustic feature vector, denoted as [e.g., ]. ,in, For acoustic feature dimensions.

[0025] S2.2 Thermal Feature Extraction: Based on data from a temperature sensor network; Calculate the maximum temperature difference or average temperature gradient between different regions on the battery surface, and express it as the spatial gradient of the temperature field. In the measurement window Within the range, calculate the fastest rate of temperature rise among all sensors, and express it as the maximum rate of temperature rise. ; Calculate all Temperature values ​​at each measuring point The standard deviation of is expressed as the temperature non-uniformity index. ,Right now ,in, Average temperature; These parameters constitute the thermal characteristic vector. This reflects the spatial heterogeneity of the generation and dissipation of internal reaction heat in the battery, and is associated with uneven degradation such as local overcharging and lithium plating. S2.3 Electrochemical Feature Extraction: Extracting the measured impedance spectrum Perform equivalent circuit model fitting or direct feature point reading; The real intercept of the high-frequency impedance is expressed as the ohmic internal resistance. The diameter of the capacitive arc in the mid-frequency region is expressed as the charge transfer resistance. The fitting parameters for the low-frequency oblique region are expressed as the Warburg diffusion coefficient. The impedance phase angle at one or more specific frequencies (e.g., 1Hz, 10Hz) is represented as the phase angle at the characteristic frequency point. ; These parameters constitute the electrochemical eigenvectors. It directly reflects the electrochemical reaction kinetics and mass transfer processes inside the battery; S2.4 Feature fusion, combining the same timestamp Below, and acoustic feature vectors associated with spatial coordinates Thermal eigenvectors and electrochemical eigenvectors (where superscript) Indicates the first splicing and aligning (associated spatial locations or regions); For example, for a specific region defined on the battery surface, the local acoustic characteristics of that region, its temperature and surrounding temperature gradient, and the overall electrochemical characteristics (or the local impedance characteristics measured by multiple electrodes) are combined to form a multidimensional feature matrix where rows represent different spatial locations / feature types and columns represent different feature dimensions. ,in, The total number of spatial locations or feature types is represented by this matrix, which is the acoustic-thermal-electric coupling feature spectrum defined in this invention. This feature spectrum is a holographic digital fingerprint of the battery's state at the current moment.

[0026] S3. Based on the acoustic-thermal-electric coupling characteristic spectrum, the degradation mechanism inside the battery is decoupled and quantitatively analyzed to obtain a quantitative index of at least one degradation mechanism. This step aims to extract the specific degradation mechanism and its contribution from the coupled characteristic spectrum, as shown in the appendix. Figure 2 As shown, the details are as follows: S3.1 Establish a standard characteristic spectrum template library. Under laboratory conditions, design a series of accelerated aging experiments to induce battery samples with different degradation modes dominated by solid electrolyte interfacial film growth, lithium metal deposition, active material loss, and electrode material particle crack propagation. For each sample, perform steps S1 and S2 periodically during its aging process, collect and extract standard coupled characteristic spectra of various degradation modes at different degradation stages, and construct a standard characteristic spectrum template library. Each template It is associated with a specific recession mechanism; S3.2 Online matching decomposition and quantization: In online applications, the battery feature spectrum matrix is ​​constructed in real time. With template library Matching can be performed using algorithms such as sparse coding or nonnegative matrix factorization. Approximate decomposition into a linear combination of standard templates: , in, That is, the corresponding number The contribution coefficient of a recession mechanism template, i.e., a quantitative indicator, for example... If it is a lithium plating template, then This can be called the lithium plating index. If it is a crack template, then These can be called crack indices, and these coefficients typically satisfy... and (After normalization), by tracking these indices in real time As time evolves, dynamic and quantitative analysis of the dominant degradation mechanisms within the battery can be achieved.

[0027] S4. Utilizing quantitative indicators of the degradation mechanism, and combining mechanistic models with data-driven models, the battery's health state is collaboratively estimated using an information fusion algorithm. This step integrates mechanistic knowledge with data-driven approaches to achieve robust estimation, as shown in the appendix. Figure 3 As shown, the details are as follows: S4.1, Mechanism Model Branch: Establish a simplified, parameterized battery electrochemical-mechanical coupled degradation model. The state variables of this model... In addition to lithium ion concentration and potential, it also specifically introduces the internal states corresponding to the analysis results of step S3, such as the amount of lithium deposited. and crack density The state-space equations of this model can be described as follows: , , in, and For state transition and observation functions, Input (current, ambient temperature). For model parameters, and For process noise and observation noise, For the voltage or capacity predicted by the model; Quantitative indicators obtained through online analysis (such as the lithium plating index) ) as an internal state For some observations or constraints, the state vector is dynamically updated through a state observer (such as an extended Kalman filter). and its error covariance matrix After the update, the health status (defined as the current maximum available capacity) is extracted from the model. With rated capacity The ratio of the two values ​​is used to obtain the first SOH estimate. Meanwhile, the filtering algorithm will provide the error covariance of this estimate. (First uncertainty); S4.2, Data-driven model branch, which uses the coupled feature spectrum matrix constructed in step S2. (or perform principal component analysis (PCA) dimensionality reduction on the principal components) ,in, much smaller Using these as input features, a machine learning model pre-trained on a historical aging dataset is employed. (Such as gradient boosting trees, lightweight neural networks) directly map to the SOH value: , in, The parameters trained for the model, To estimate the prediction error, the uncertainty of the prediction is estimated through cross-validation or the model's own probability output, thus obtaining a second SOH estimate. and the model's own prediction variance (Second uncertainty); S4.3 Adaptive weighted fusion: Covariance-weighted fusion (or equivalent to the information fusion formula in Bayesian estimation) is used to optimally fuse the estimation results of the two independent branches, resulting in the final SOH estimate. and the covariance after fusion The calculation is as follows: , , Formula parameter description: and These are referred to as the information matrix or accuracy matrix of the mechanistic model and the data-driven model, respectively. The larger the inverse covariance (accuracy), the more accurate the estimate (the smaller the uncertainty). Covariance after fusion It is the inverse of the sum of the information matrices of the two, which means that the fused estimate incorporates the information from both sides, and its uncertainty must be less than or equal to the uncertainty of any single estimate. Final SOH estimate It is a weighted average of two estimates, with the weights proportional to their respective accuracy matrices (inverse covariance). Its physical meaning is that at any given time, the model whose estimate is more reliable (has less uncertainty) will have a greater weight in the final result. This dynamic weighting strategy cleverly combines the extrapolation robustness of mechanistic models with the advantages of data-driven models in fitting complex relationships.

[0028] To verify the effectiveness of this invention, a cycle aging experiment was conducted on a group of commercial NMC ternary lithium-ion power batteries, and the results were compared with two traditional methods (based on the ICA characteristics of the charging curve and the SOH estimation method based on the internal resistance of the equivalent circuit model). The experimental results show that: In terms of mechanism analysis: When the capacity decays to 95% of the initial capacity, the lithium plating index of the method of the present invention shows a continuous upward trend, which successfully warns of the accelerated capacity decay caused by lithium plating. In contrast, neither of the two traditional methods found any obvious abnormal characteristics at this stage. Regarding SOH estimation: Throughout the entire aging cycle (capacity decay to 80%), the mean absolute error (MAE) of SOH estimation by the method of this invention (after fusion) is less than 1.0%, and its 95% confidence interval can reliably cover the true capacity decay trajectory. In contrast, the MAE of the method based on ICA features is about 2.5%, and the error increases significantly in the second half of the cycle. The MAE of the method based on internal resistance exceeds 3.0%, and the estimation results fluctuate greatly.

[0029] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for analyzing the degradation mechanism and estimating the health status of a power battery, characterized in that: Specifically, the following steps are included: S1. During battery operation, multi-physics field in-situ sensing signals are collected synchronously, including at least actively excited ultrasonic scanning signals, distributed temperature sensing signals, and electrochemical impedance spectroscopy signals. S2. Characteristic parameters that can characterize the internal mechanical structure, thermal behavior and electrochemical reaction kinetics of the battery are extracted from ultrasonic scanning signals, distributed temperature sensing signals and electrochemical impedance spectroscopy signals respectively, and then fused to construct a spatiotemporally correlated acoustic-thermal-electric coupling characteristic spectrum. S3. Based on the acoustic-thermal-electric coupling characteristic spectrum, the degradation mechanism inside the battery is decoupled and quantitatively analyzed to obtain a quantitative index of at least one degradation mechanism. S4. Using quantitative indicators of degradation mechanisms, combined with mechanism models and data-driven models, the health status of the battery is estimated collaboratively through information fusion algorithms.

2. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 1, characterized in that: In step S1, the synchronous acquisition specifically involves: during the battery's resting phase or the voltage plateau phase of constant current charging, triggering a collaborative measurement cycle to control the ultrasonic excitation unit, impedance excitation unit, and temperature acquisition unit to perform millisecond-level time-synchronized signal acquisition.

3. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 1, characterized in that: In step S2, extracting feature parameters from the ultrasonic scanning signal includes: processing the received ultrasonic echo signal and extracting acoustic feature vectors related to changes in the internal microstructure of the battery. The acoustic feature vectors include at least one or more of the following: sound wave propagation speed through each layer of the battery, signal energy attenuation coefficient, spectral amplitude of a specific frequency band, or scattering intensity.

4. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 1, characterized in that: In step S2, extracting feature parameters from the distributed temperature sensing signal includes: calculating a thermal feature vector based on temperature sensor readings arranged at multiple locations on the battery surface, wherein the thermal feature vector includes at least one or more of the following: temperature field spatial gradient, maximum temperature rise rate, and temperature distribution non-uniformity index.

5. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 1, characterized in that: In step S2, extracting characteristic parameters from the electrochemical impedance spectroscopy signal includes: analyzing the measured impedance spectrum and extracting electrochemical feature vectors, wherein the electrochemical feature vectors include at least one or more of the following: ohmic internal resistance, charge transfer resistance, Warburg diffusion coefficient, and phase angle at one or more characteristic frequency points.

6. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 1, characterized in that: In step S2, the specific method for fusing and constructing the acoustic-thermal-electric coupling feature spectrum is as follows: the acoustic feature vector, thermal feature vector and electrochemical feature vector at the same time and spatially related are spliced ​​and aligned to form a multi-dimensional feature matrix, which serves as the coupling feature spectrum characterizing the overall state of the battery at that moment.

7. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 1, characterized in that: Step S3 specifically includes: S3.

1. Establish a standard characteristic spectrum template library corresponding to different dominant decay mechanisms in advance through experiments; S3.

2. Match and decompose the acoustic-thermal-electric coupling characteristic spectrum constructed online with the standard characteristic spectrum template library, and calculate the contribution coefficient of each decay mechanism template to the current battery state. The contribution coefficient is the quantitative index of the decay mechanism.

8. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 7, characterized in that: The dominant degradation mechanism includes at least two of the following: solid electrolyte interfacial film growth, lithium metal deposition, loss of active material, and particle crack propagation in electrode material. The quantitative indicators include the lithium deposition index for characterizing the severity of lithium deposition and / or the crack index for characterizing the propagation of structural cracks.

9. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 1, characterized in that: Step S4 specifically includes: S4.1 Establish a battery electrochemical-mechanical coupled degradation model that includes quantitative indicators of degradation mechanism as state variables, update the model using quantitative indicators, and obtain the first health state estimate and the first uncertainty based on the mechanism model. S4.2 Using the acoustic-thermal-electric coupling characteristic spectrum or its principal components after dimensionality reduction as input, the second health state estimate and the second uncertainty are obtained through a data-driven model; S4.3 Based on the first uncertainty and the second uncertainty, an adaptive weighted fusion algorithm is used to fuse the first health state estimate and the second health state estimate, and output the final battery health state estimate and confidence interval.

10. The method for analyzing the degradation mechanism and estimating the health status of a power battery according to claim 9, characterized in that: In step S4.3, the adaptive weighted fusion algorithm is a fusion framework based on Kalman filtering or Bayesian estimation, and its weights are dynamically adjusted according to the estimation error covariance of each model under the current operating conditions.

Images