Solid-state battery performance test method and system based on data analysis

Abstract

The invention discloses a solid-state battery performance testing method and system based on data analysis, and relates to the field of battery performance testing, and the method comprises the steps: building a multi-physics field coupled digital twinborn model, collecting a strain signal and temperature field distribution of a failure region according to a three-dimensional physics field state map of the failure region, and when abnormally abruptly changed, carrying out the testing of the performance of a solid-state battery. Triggering multi-stage early warning, matching physical failure evidences with a historical database, extracting an optimal test parameter combination through a meta-learning framework, dynamically adjusting a charging and discharging strategy and monitoring frequency, generating a real-time data stream, and updating parameters of a multi-physics field coupled digital twin model through a back propagation algorithm. According to the method, the test parameters are dynamically optimized through the meta-learning framework, the performance of the solid-state battery is monitored and predicted, the safety and the reliability of the solid-state battery are improved, and a charging and discharging strategy is optimized to prolong the service life of the battery.

Description

Technical Field

[0001] The present invention relates to the field of battery performance testing, and in particular to a solid-state battery performance testing method and system based on data analysis. Background Art

[0002] As a new type of high-performance battery technology, solid-state batteries have received widespread attention in the new energy field in recent years. With the rapid development of electric vehicles, portable electronic devices and large-scale energy storage applications, the requirements for battery energy density, safety, cycle life and fast charging and discharging capabilities are becoming increasingly higher. Solid-state batteries use solid electrolytes instead of traditional liquid electrolytes, and have higher ionic conductivity, more stable chemical properties and lower interfacial impedance, thus showing significant performance advantages.

[0003] Although there are currently a variety of technical means for solid-state battery performance testing, existing technologies are limited to the measurement of a single physical field and cannot fully reflect the complex multi-physical field coupling effects within solid-state batteries. In addition, traditional testing methods usually rely on fixed experimental conditions, lack the ability to dynamically respond to real-time data streams, and cannot self-adjust according to the actual working status of the battery. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a solid-state battery performance testing method based on data analysis to solve the problem that the existing technology is limited to the measurement of a single physical field and is difficult to fully reflect the complex multi-physical field coupling effects inside the solid-state battery.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a solid-state battery performance testing method based on data analysis, which includes: establishing a multi-physics field coupled digital twin model based on the material parameters and structural parameters of the target solid-state battery, and outputting a three-dimensional physical field state map of the failure area; Based on the three-dimensional physical field state map of the failure area, the strain signal and temperature field distribution of the failure area are collected. When abnormal mutations are detected, multi-level warnings are triggered and physical failure evidence is obtained; Matching physical failure evidence with historical databases, extracting optimal test parameter combinations through a meta-learning framework, dynamically adjusting charge and discharge strategies and monitoring frequency, executing the adjusted test process, and generating real-time data streams; Based on real-time data streams, the back-propagation algorithm is used to update the parameters of the multi-physics coupled digital twin model and output the revised multi-physics coupled digital twin model. Based on the revised multi-physics field coupled digital twin model, the timing sequence of the solid-state battery is analyzed in real time, the failure probability curve is obtained, and a test report is generated.

[0007] As a preferred embodiment of the solid-state battery performance testing method based on data analysis of the present invention, the material parameters include the composition of the positive and negative electrode active materials, the type of solid electrolyte, and the ratio of the binder to the conductive agent; The structural parameters include electrode thickness, porosity, particle size distribution, current collector type and interface layer thickness.

[0008] As a preferred solution of the solid-state battery performance testing method based on data analysis described in the present invention, wherein: the digital twin model of multi-physical field coupling is established to output a three-dimensional physical field state map of the failure area, the steps are as follows: Based on the material and structural parameters of the target solid-state battery, synchrotron radiation X-ray tomography is used to reconstruct the three-dimensional voxel matrix inside the electrode with nanometer resolution and obtain the three-dimensional microstructure of the electrode; Based on the three-dimensional voxel matrix, the three-dimensional microstructure of the electrode is discretized into regular grid units. Material parameters and structural parameters are injected into each grid unit to establish the electrochemical-thermal-mechanical coupling equation; Based on the electrochemical-thermomechanical coupling equations, a multi-physics field coupling digital twin model is constructed to dynamically reflect the ion concentration gradient distribution, stress field distribution, and temperature field distribution; Based on the three-dimensional microstructure of the electrode provided by the three-dimensional voxel matrix, the digital twin model coupled with multiple physical fields dynamically outputs the three-dimensional physical field state map of ion concentration gradient distribution, stress distribution field and temperature field distribution; Through the regional anomaly frequency in historical failure cases, the failure area positions of the lithium dendrite growth zone and the interface peeling zone are dynamically identified in the three-dimensional physical field state map, forming a three-dimensional physical field state map of the failure area.

[0009] As a preferred embodiment of the solid-state battery performance testing method based on data analysis described in the present invention, wherein: the strain signal and temperature field distribution of the failure area are collected, and when an abnormal mutation is detected, a multi-level warning is triggered and physical failure evidence is obtained. The steps are as follows: Based on the three-dimensional physical field state map of the failure area, distributed optical fiber strain sensors and non-contact infrared thermal imagers are installed on the surface of the dynamically identified lithium dendrite growth area and interface peeling area to collect local strain signals and surface temperature field distribution in real time; The fiber optic strain sensor continuously outputs strain time series data. When a sudden increase in local strain is detected, it is determined to be the critical point of mechanical failure. The non-contact infrared thermal imager outputs a temperature field distribution map. When the local temperature suddenly changes abnormally, it is determined to be a precursor to thermal runaway. When any of the mechanical failure critical point and thermal runaway precursors occur, multi-level warnings are immediately triggered and a warning record list is obtained; According to the multi-level early warning and safety control, when the early warning is triggered, the physical failure evidence of the lithium dendrite growth area and the interface peeling area is collected through field emission scanning electron microscopy and X-ray photoelectron spectroscopy respectively.

[0010] As a preferred embodiment of the data analysis-based solid-state battery performance testing method of the present invention, the steps of matching physical failure evidence with a historical database, extracting the optimal test parameter combination through a meta-learning framework, dynamically adjusting the charge and discharge strategy and monitoring frequency, executing the adjusted test process, and generating a real-time data stream are as follows: The historical database includes crack morphology characteristics, monitoring frequency values, decomposition product components, charge and discharge rate values, pressure loading values and unit cycle failure event frequency; The feature vectorization method is used to convert physical failure evidence and historical database into multidimensional vectors and historical database vectors. The matching degree between the multidimensional vectors and the historical database vectors is calculated by cosine similarity to screen the failure case subset. Based on a subset of failure cases, a radial basis kernel covariance function is used to construct a covariance matrix. The objective function is defined as the negative log marginal likelihood. The cycle life and polarization voltage of the solid-state battery are used as constraints to construct a Gaussian process regression surrogate model. The UE learning framework is used to initialize the search space for charge / discharge rate and pressure loading values. The Gaussian regression proxy model is then used to predict the probability distribution of the charge / discharge rate and pressure loading values. The expected improvement criterion is then used to select the optimal solution that satisfies the constraints and obtain the optimal test parameter combination. According to the predicted charge and discharge rate value in the optimal test parameter combination, the current curve shape of the constant current charge and discharge test is adjusted, and the monitoring frequency value is adjusted according to the pressure loading value to form a test parameter log after adjustment; After the segmented constant current mode and monitoring frequency adjustment are completed, the constant current charge and discharge test is performed through the battery testing method, combined with the DC internal resistance method to calculate the internal resistance and the infrared thermal imager to monitor the temperature, generating a real-time data stream.

[0011] As a preferred solution of the solid-state battery performance testing method based on data analysis described in the present invention, wherein: the parameters of the multi-physics field coupled digital twin model are updated by the back propagation algorithm, and the revised multi-physics field coupled digital twin model is output. The specific steps are as follows: The real-time data stream is input into the Levenberg-Marquardt optimizer. By minimizing the residual between the actual observation data and the theoretical prediction value of the multi-physics field coupled digital twin model, a back propagation algorithm is implemented to correct the solid phase diffusion coefficient and interface reaction term in the multi-physics field coupled digital twin model. Combined with the actual cycle capacity attenuation data in the real-time data stream, the degradation equation of the multi-physics field coupled digital twin model is reversely fitted to calibrate the nonlinear cumulative effect of the active material loss rate during the cycle process, and output the corrected multi-physics field coupled digital twin model.

[0012] As a preferred solution of the solid-state battery performance testing method based on data analysis described in the present invention, wherein: the real-time analysis of the solid-state battery timing sequence, and the acquisition of the failure probability curve, and the generation of the test report are carried out, the specific steps are as follows: Dynamically simulate the charge and discharge cycle process based on the modified multi-physics field coupled digital twin model, analyze the real-time generated time series of voltage, current and temperature, and extract the voltage polarization curve shape and capacity decay rate value; Based on the unit cycle failure event frequency in the historical database, the adjacent difference values of the time series are calculated in real time as the decay rate slope to form the input data set for the Granger causality test; Based on the voltage polarization curve morphology and capacity decay rate value, Granger causality test is used to correlate the capacity decay rate slope with the failure event frequency, generate the cycle number-failure probability curve, and mark the critical threshold; Based on the cycle number-failure probability curve, the 3D physical field state map of the failure area, the multi-level warning record and the adjusted test parameter log, the MCP protocol is used to intelligently align the standardized template with multi-source data to generate a test report.

[0013] In a second aspect, the present invention provides a solid-state battery performance testing system based on data analysis, comprising: Failure module, early warning module, test module, correction module and analysis module, The failure module establishes a multi-physics field coupled digital twin model based on the material and structural parameters of the target solid-state battery, and outputs a three-dimensional physical field state map of the failure area; The early warning module collects strain signals and temperature field distribution in the failure area based on the three-dimensional physical field state map of the failure area. When abnormal mutations are detected, multi-level early warnings are triggered and physical failure evidence is obtained; The test module matches physical failure evidence with a historical database, extracts the optimal test parameter combination through a meta-learning framework, dynamically adjusts the charge and discharge strategy and monitoring frequency, executes the adjusted test process, and generates a real-time data stream; The correction module updates the parameters of the multi-physics field coupled digital twin model through the back-propagation algorithm based on the real-time data stream and outputs the corrected multi-physics field coupled digital twin model; The analysis module, based on the revised multi-physics field coupled digital twin model, analyzes the timing sequence of the solid-state battery in real time, obtains the failure probability curve, and generates a test report.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the solid-state battery performance testing method based on data analysis as described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the solid-state battery performance testing method based on data analysis as described in the first aspect of the present invention.

[0016] The beneficial effects of the present invention are as follows: the present invention establishes a digital twin model of multi-physical field coupling, collects the strain signal and temperature field distribution of the solid-state battery in real time, and combines the meta-learning framework to dynamically optimize the test parameters, thereby realizing the monitoring and prediction of the solid-state battery performance, helping to improve the safety and reliability of the solid-state battery, and optimizing the charging and discharging strategy to extend the battery life. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of 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 paying any creative work.

[0018] Figure 1 Flowchart of the solid-state battery performance testing method based on data analysis.

[0019] Figure 2 Build a flow chart for a multiphysics coupled digital twin model.

[0020] Figure 3 Generate and modify flow charts for real-time data streams.

[0021] Figure 4 Generate flow charts for failure probability analysis and test reports. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0025] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides a solid-state battery performance testing method based on data analysis, comprising the following steps: S1. Based on the material parameters and structural parameters of the target solid-state battery, a multi-physics field coupled digital twin model is established to output a three-dimensional physical field state map of the failure area; The specific steps are as follows: Material parameters include the composition of positive and negative electrode active materials, solid electrolyte type, and the ratio of binder to conductive agent; structural parameters include electrode thickness, porosity, particle size distribution, current collector type, and interface layer thickness.

[0026] Based on the material and structural parameters of the target solid-state battery, synchrotron X-ray tomography is used to reconstruct the three-dimensional voxel matrix inside the electrode with nanometer resolution, and obtain the three-dimensional microstructure of the electrode, including the spatial distribution and connectivity of the active material phase, pore phase, and electrolyte phase; It should be noted that based on the material parameters and structural parameters of the target solid-state battery, synchrotron X-ray tomography is used to rotate and scan the battery electrodes with nanometer resolution using a parallel beam of hard X-rays, and projection images at different angles are collected. The projection images are arranged in order according to the rotation angle to form a two-dimensional projection sequence; the two-dimensional projection sequence is converted into a three-dimensional voxel matrix through a filtered back-projection algorithm, and the voxel grayscale value reflects the difference in the linear attenuation coefficient of the material to X-rays; the active material phase, pore phase and electrolyte phase are distinguished through a threshold segmentation algorithm, and the spatial distribution and connectivity parameters of the active material phase, pore phase and electrolyte phase are quantified in combination with morphological operations, and finally the three-dimensional microstructure of the electrode including the phase interface contact area and ion transmission path is output.

[0027] Based on the three-dimensional voxel matrix, the three-dimensional microstructure of the electrode is discretized into regular grid units. Material parameters and structural parameters are injected into each grid unit to establish the electrochemical-thermal-mechanical coupling equation; It should be noted that based on the three-dimensional voxel matrix, the octree space partitioning algorithm is used to discretize the three-dimensional microstructure of the electrode into hexahedral regular grid cells, and the material parameters and structural parameters are injected into each grid cell. In COMSOL Multiphysics, the finite volume method is used to fully couple the electrochemical, thermal, and mechanical equations discretized into the hexahedral regular grid cells. Among them, the electrochemical equation is coupled with the thermal equation through the Joule heat term, the thermal equation is coupled with the mechanical equation through the thermal expansion term, and the mechanical equation is coupled with the electrochemical equation through the stress-dependent interface reaction rate term, thus constructing a set of fully coupled electrochemical-thermal-mechanical control equations. Based on the electrochemical-thermomechanical coupling equations, a multi-physics field coupling digital twin model is constructed to dynamically reflect the ion concentration gradient distribution, stress field distribution, and temperature field distribution; The steps to build a multi-physics field coupled digital twin model are as follows: The electrochemical equation is constructed based on the ion diffusion characteristics in the material parameters, and the lithium ion concentration gradient distribution is calculated using the Nernst-Planck equation, where the solid phase diffusion coefficient is determined by the active material composition in the material parameters; The thermal equation is constructed by combining Fourier's heat conduction law with the porosity in the structural parameters to calculate the temperature field distribution formed by the local heat generation rate. The thermal conductivity is determined by the electrolyte type and the binder ratio. The mechanical equation is constructed based on the particle size distribution of active particles. The linear elastic mechanics equation is used to quantify the electrochemical expansion stress. The resulting stress distribution is constrained by the particle size and collector type.

[0028] Based on the three-dimensional microstructure of the electrode provided by the three-dimensional voxel matrix, the digital twin model coupled with multiple physical fields dynamically outputs the three-dimensional physical field state map of ion concentration gradient distribution, stress distribution field and temperature field distribution; By using the regional anomaly frequency in historical failure cases, the failure area locations of the lithium dendrite growth zone and the interface peeling zone are dynamically identified in the three-dimensional physical field state map, forming a three-dimensional physical field state map of the failure area; It should be noted that based on the ion concentration gradient distribution, stress distribution and temperature field distribution in the three-dimensional physical field state map, the physical characteristic criterion of the lithium dendrite growth zone is first extracted: when the local ion concentration gradient exceeds the critical value and the current density exceeds the current threshold, the current threshold is defined according to the space charge theory. For example, the current threshold value range is [0.8-1.2] mA / cm 2At the same time, combined with the regional abnormal frequency data in the historical failure case library, the spatial failure probability distribution is calculated by Gaussian kernel density estimation, and the real-time physical field data and the failure probability distribution are Bayesian fused to dynamically identify the failure area that meets the joint conditions, and generate a three-dimensional physical field state map of the failure area.

[0029] S2. Based on the three-dimensional physical field state map of the failure area, the strain signal and temperature field distribution of the failure area are collected. When abnormal mutations are detected, multi-level warnings are triggered and physical failure evidence is obtained; The specific steps are: Based on the three-dimensional physical field state map of the failure area, distributed optical fiber strain sensors and non-contact infrared thermal imagers are installed on the surface of the dynamically identified lithium dendrite growth area and interface peeling area to collect local strain signals and surface temperature field distribution in real time; It should be explained that in the three-dimensional physical field state map of the failure area, distributed optical fiber strain sensors are attached to the surface of the lithium dendrite growth area and the interface peeling area in the form of an array, and the local strain signal of the lithium dendrite growth area caused by the deposition of metallic lithium is measured; the non-contact infrared thermal imager synchronously collects the surface temperature field distribution of the interface peeling area, and locates the hot spot area based on the thermal anomaly threshold. Among them, the thermal anomaly threshold is based on the empirical judgment in the battery thermal runaway warning, and is used to distinguish normal heat diffusion from abnormal local overheating.

[0030] The non-contact infrared thermal imager locates the abnormal temperature gradient area by detecting the surface temperature field distribution in the interface peeling area, and aligns the timestamps with the abnormal stress signal in the lithium dendrite growth area captured by the distributed optical fiber strain sensor to achieve real-time monitoring of the electrochemical-thermal-mechanical dual-field coupling.

[0031] The fiber optic strain sensor continuously outputs strain time series data. When a sudden increase in local strain is detected, it is determined to be the critical point of mechanical failure. The non-contact infrared thermal imager outputs a temperature field distribution map. When the local temperature suddenly changes abnormally, it is determined to be a precursor to thermal runaway. When mechanical failure critical points and thermal runaway precursors appear, multi-level warnings are immediately triggered and a list of warning records is obtained; Among them, the multi-level warning includes level one sound and light alarm, level two charging and discharging suspension, and level three safety isolation.

[0032] According to the multi-level warning, when the warning is triggered, field emission scanning electron microscopy and X-ray photoelectron spectroscopy are used to collect physical failure evidence in the interface debonding area and lithium dendrite growth area respectively; It should be noted that after triggering the multi-level warning, a field emission scanning electron microscope is used to perform a low-voltage scan of the lithium dendrite growth area, and secondary electron imaging is used to obtain the three-dimensional morphology and stress concentration area of the dendrite growth area. At the same time, X-ray photoelectron spectroscopy uses Al Kα rays to excite photoelectrons in the interface debonding area, and analyzes the binding energy shift to detect the chemical state changes of the electrolyte decomposition products. After the three-dimensional morphology of the lithium dendrite growth area scanned by field emission scanning electron microscopy and X-ray photoelectron spectroscopy and the photoelectrons in the stress concentration area and interface debonding area were aligned with the timestamps, the three-dimensional morphology of the lithium dendrite growth area scanned by field emission scanning electron microscopy and the crack propagation path in the stress concentration area and the element valence state shift detected by X-ray photoelectron spectroscopy through Al Kα ray excitation of the interface debonding area and analysis of the binding energy shift together constitute physical failure evidence.

[0033] S3, matching physical failure evidence with historical databases, extracting optimal test parameter combinations through a meta-learning framework, dynamically adjusting the charge and discharge strategy and monitoring frequency, executing the adjusted test process, and generating real-time data streams; The specific steps are as follows: The historical database includes crack morphology characteristics, monitoring frequency values, decomposition product composition, charge and discharge rate values, pressure loading values, and failure event frequency per unit cycle; The feature vectorization method is used to convert physical failure evidence and historical database into multidimensional vectors and historical database vectors. The matching degree between the multidimensional vectors and the historical database vectors is calculated by cosine similarity to screen the failure case subset. It should be noted that the physical failure evidence is converted into a 128-dimensional feature vector after dimensionality reduction through principal component analysis, and the cases in the historical database are processed with the same PCA parameters to generate the historical database vector; and the cosine similarity is used to calculate the matching degree between the physical failure evidence feature vector and the historical database vector. , the formula is: ; in, Expressed as the physical failure evidence feature vector in the The component value of the dimension, Represented as the historical database vector in The component value of the dimension; The K-nearest neighbor algorithm is used to sort the physical failure evidence feature vectors and the historical database vectors according to their cosine similarity, and the top K cases with the highest similarity are directly selected as the matching subset. Among them, the failure case subset includes cycle life, polarization voltage, material parameters and structural parameters.

[0034] Based on a subset of failure cases, a radial basis kernel covariance function is used to construct a covariance matrix. The objective function is defined as the negative log marginal likelihood. The cycle life and polarization voltage of the solid-state battery are used as constraints to construct a Gaussian process regression surrogate model. It should be noted that the cycle life and polarization voltage in the failure case subset were Z-score standardized, and the maximum value of the ion concentration gradient distribution, the stress concentration coefficient and the temperature field mutation amplitude were extracted as input features, and the cycle life and polarization voltage were used as output targets. Subsequently, the radial basis kernel covariance function was used to describe the nonlinear relationship between the input features, and the covariance matrix was constructed. The objective function was defined as the negative log marginal likelihood. Then, the lower limit of the cycle life and the upper limit of the polarization voltage of the solid-state battery were set as constraints, and the conjugate gradient method was used to optimize the hyperparameters of the radial basis kernel covariance function to obtain a Gaussian process regression proxy model.

[0035] The UE learning framework is used to initialize the search space for charge / discharge rate and pressure loading values. The Gaussian regression proxy model is then used to predict the probability distribution of the charge / discharge rate and pressure loading values. The expected improvement criterion is then used to select the optimal solution that satisfies the constraints and obtain the optimal test parameter combination. It should be noted that under the framework of uMe learning, within the range of charge and discharge rate values and pressure loading values, the initial parameter combination is generated by Latin hypercube sampling, and the mapping relationship between the initial parameter combination and the battery capacity attenuation rate is established using the Gaussian process regression model. The output is a predictive distribution containing the mean and variance, and the potential improvement space of the new parameter combination compared to the current optimal solution is quantified by the expected improvement criterion. The Pareto frontier is screened in combination with the constraints of cycle life and polarization voltage to always obtain the optimal test parameter combination.

[0036] According to the predicted charge and discharge rate value in the optimal test parameter combination, the current curve shape of the constant current charge and discharge test is adjusted, and the monitoring frequency value is adjusted according to the pressure loading value to form a test parameter log after adjustment; It should be noted that according to the charge and discharge rate value predicted in the optimal test parameter combination, the current curve shape of the constant current charge and discharge test is adjusted to a segmented constant current mode, and the initial current is used for charging in the charging stage. When the voltage reaches the upper limit, it is switched to constant current charging until the cut-off voltage; Based on the dynamic range of the pressure loading value, the monitoring frequency value is dynamically adjusted according to a linear relationship, and the current, voltage and pressure time series data are fully recorded during the charge and discharge cycle to generate an adjusted test parameter log containing timestamp, charge and discharge rate value, pressure loading value, operating condition parameters, monitoring frequency value and corresponding performance indicators.

[0037] After the segmented constant current mode and monitoring frequency adjustment are completed, the constant current charge and discharge test is performed through the battery testing method, combined with the DC internal resistance method to calculate the internal resistance and the infrared thermal imager to monitor the temperature, generating a real-time data stream.

[0038] It should be noted that after the charge and discharge strategy and monitoring frequency are adjusted, constant current charge and discharge tests are performed using the battery testing method, and the battery internal resistance is calculated in real time using the DC internal resistance method. At the same time, the infrared thermal imager collects temperature distribution data at the adjusted monitoring frequency value to generate a real-time data stream containing timestamps, current, voltage, internal resistance, temperature, pressure loading values, minimized actual observation data, and actual cycle capacity attenuation data.

[0039] Among them, minimizing the actual observed data is obtained by predicting the capacity attenuation rate through a Gaussian process regression proxy model, and then optimizing the residual sum of squares between the theoretical value and the measured value using the least squares method; the actual cycle capacity attenuation data is calculated by the ratio of the discharge capacity to the initial capacity recorded in the constant current charge and discharge test, and is generated based on dynamic calibration of the temperature and rate correction coefficient.

[0040] S4. Based on the real-time data stream, the parameters of the multi-physics field coupled digital twin model are updated through the back-propagation algorithm, and the revised multi-physics field coupled digital twin model is output; The specific steps are as follows: The real-time data stream is input into the Levenberg-Marquardt optimizer. By minimizing the residual between the actual observation data and the theoretical prediction value of the multi-physics field coupled digital twin model, a back propagation algorithm is implemented to correct the solid phase diffusion coefficient and interface reaction term in the multi-physics field coupled digital twin model. It should be noted that after the real-time data stream is input into the Levenberg-Marquardt optimizer, the objective function is constructed by calculating the residual vector between the predicted value of the multi-physics field coupled digital twin model and the actual observation data, and the Jacobian matrix is used to iteratively solve the parameter correction amount according to the Levenberg-Marquardt update rule. Among them, the parameter correction amount is obtained by comparing the residual between the model predicted value and the actual observation data, and the Levenberg-Marquardt optimizer is used to dynamically adjust the damping factor and Jacobian matrix, and iteratively optimize the adjustment values of the solid phase diffusion coefficient and interface reaction term of the multi-physics field digital twin model until the residual converges.

[0041] Combined with the actual cycle capacity decay data in the real-time data stream, the degradation equation of the multi-physics field coupled digital twin model is back-fitted to calibrate the nonlinear cumulative effect of the active material loss rate during the cycle process, and the corrected multi-physics field coupled digital twin model is output; The active material loss rate is determined by calibrating the nonlinear cumulative effect parameters during the cycle using the degradation equation.

[0042] It should be noted that the actual cycle capacity attenuation data in the real-time data stream is input into the multi-physics field coupled digital twin model, the residual function is constructed by reverse fitting the degradation equation, and the Levenberg-Marquardt optimizer is used to iteratively solve the nonlinear cumulative effect parameters of the active material loss rate value. After each iteration, the coupling terms of the solid phase diffusion coefficient and the interface reaction term rate in the multi-physics field coupled digital twin model are corrected by the back propagation algorithm until the residual converges, and the corrected multi-physics field coupled digital twin model containing the calibrated active material loss rate equation and multi-field coupling parameters is output.

[0043] S5. Based on the modified multi-physics field coupled digital twin model, the solid-state battery timing sequence is analyzed in real time, and the failure probability curve is obtained to generate a test report; The specific steps are as follows: Dynamically simulate the charge and discharge cycle process based on the modified multi-physics field coupled digital twin model, analyze the real-time generated time series of voltage, current and temperature, and extract the voltage polarization curve shape and capacity decay rate value; It should be noted that when dynamically simulating the charge and discharge cycle process based on the revised multi-physics field coupling digital twin model, the real-time collected voltage, current and temperature time series are first input into the multi-physics field coupling digital twin model as boundary conditions, and the lithium ion concentration gradient distribution is obtained by solving the Nernst-Planck equation. The temperature field distribution is calculated in combination with Fourier's heat conduction law, and the stress distribution is updated based on the linear elastic mechanics equation. Subsequently, the voltage, current and temperature time series are analyzed and the voltage polarization curve morphology and capacity attenuation rate value are extracted, where the attenuation rate constant is extracted by the capacity increment analysis method.

[0044] Based on the unit cycle failure event frequency in the historical database, the adjacent difference values of the time series are calculated in real time as the decay rate slope to form the input data set for the Granger causality test; It should be noted that based on the unit cycle failure event frequency in the historical database, the failure event frequency time series is first extracted, and the difference between adjacent time points is calculated through differential operation to generate a decay rate slope series. Subsequently, the decay rate slope series is tested for stationarity and the lag order is selected to ensure that the prerequisites for the Granger causality test are met. The decay rate slope series is combined with the target variable time series to obtain a multivariate input data set for Granger causality test analysis. Among them, the target variable time series sequence comes from the direct pairing combination of the preprocessed attenuation rate slope sequence and the unit cycle failure event frequency time series sequence in the historical database.

[0045] Based on the voltage polarization curve morphology and capacity decay rate value, Granger causality test is used to correlate the capacity decay rate slope with the failure event frequency, generate the cycle number-failure probability curve, and mark the critical threshold; It should be noted that characteristic parameters are extracted from the morphology of the voltage polarization curve, and the changes in the capacity decay rate values at adjacent time points are calculated by the difference method to form a slope sequence; the capacity decay rate slope sequence is subjected to a Granger causality test with the time series of failure event frequencies per unit cycle in the historical database. Specifically, the capacity decay rate slope sequence is used as the independent variable, and the failure event frequency sequence is used as the dependent variable. The predictive effectiveness of the capacity decay rate slope on failure events is verified by comparing the difference in the sum of squares of the regression residuals with and without the lag term of the capacity decay slope sequence; a mapping relationship between the number of cycles and the failure probability is established through logistic regression, and a cycle number-failure probability curve is fitted and generated, and the number of cycles corresponding to the mutation point of the capacity decay rate slope is marked on the cycle number-failure probability curve as the critical threshold.

[0046] Generate a test report by intelligently aligning standardized templates with multi-source data through the MCP protocol based on the cycle-failure probability curve, the 3D physical field state map of the failure area, multi-level warning records, and adjusted test parameter logs. It should be explained that the failure probability time series data is extracted based on the cycle number-failure probability curve, and the spatial strain / stress distribution characteristics are analyzed in combination with the three-dimensional physical field state map of the failure area. The warning level and trigger timestamp in the three-level warning record are synchronously associated, and the time series is aligned with the operating condition parameters in the adjusted test parameter log. Through the structured ETL process, multi-source data cleaning, field mapping and unit standardization are completed, and finally integrated into a structured test report including a failure probability trend chart, a three-dimensional physical field isosurface rendering chart, a warning event timeline and a parameter adjustment comparison table.

[0047] This embodiment also provides a solid-state battery performance testing system based on data analysis, including: a failure module, an early warning module, a testing module, a correction module and an analysis module. The failure module establishes a multi-physics field coupled digital twin model based on the material and structural parameters of the target solid-state battery, and outputs a three-dimensional physical field state map of the failure area; The early warning module collects strain signals and temperature field distribution in the failure area based on the three-dimensional physical field state map of the failure area. When abnormal mutations are detected, multi-level early warnings are triggered and physical failure evidence is obtained; The test module matches physical failure evidence with a historical database, extracts the optimal test parameter combination through a meta-learning framework, dynamically adjusts the charge and discharge strategy and monitoring frequency, executes the adjusted test process, and generates a real-time data stream; The correction module updates the parameters of the multi-physics field coupled digital twin model through the back-propagation algorithm based on the real-time data stream and outputs the corrected multi-physics field coupled digital twin model; The analysis module analyzes the timing sequence of solid-state batteries in real time based on the modified multi-physics field coupled digital twin model, obtains the failure probability curve, and generates a test report; This embodiment also provides a computer device, which is suitable for the case of a solid-state battery performance testing method based on data analysis, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the solid-state battery performance testing method based on data analysis proposed in the above embodiment.

[0048] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0049] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the solid-state battery performance testing method based on data analysis as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0050] In summary, the present invention achieves monitoring and prediction of solid-state battery performance by: establishing a digital twin model coupled with multiple physical fields, collecting the strain signals and temperature field distribution of solid-state batteries in real time, and dynamically optimizing test parameters in combination with a meta-learning framework. This helps to improve the safety and reliability of solid-state batteries and optimize charging and discharging strategies to extend battery life.

[0051] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A solid-state battery performance testing method based on data analysis, characterized in that: include: Based on the material parameters and structural parameters of the target solid-state battery, a multi-physics field coupled digital twin model is established to output a three-dimensional physical field state map of the failure area; Based on the 3D physical field state map of the failure area, the strain signal and temperature field distribution of the failure area are collected. When abnormal mutations are detected, multi-level warnings are triggered and physical failure evidence is obtained. Matching physical failure evidence with historical databases, extracting optimal test parameter combinations through a meta-learning framework, dynamically adjusting charge and discharge strategies and monitoring frequency, executing the adjusted test process, and generating real-time data streams; Based on real-time data streams, the back-propagation algorithm is used to update the parameters of the multi-physics coupled digital twin model and output the revised multi-physics coupled digital twin model. Based on the revised multi-physics field coupled digital twin model, the timing sequence of the solid-state battery is analyzed in real time, the failure probability curve is obtained, and a test report is generated.

2. The solid-state battery performance testing method based on data analysis according to claim 1, characterized in that: The material parameters include the composition of the positive and negative active materials, the type of solid electrolyte, and the ratio of the binder to the conductive agent; The structural parameters include electrode thickness, porosity, particle size distribution, current collector type and interface layer thickness.

3. The solid-state battery performance testing method based on data analysis according to claim 1, characterized in that: The steps of establishing a multi-physics field coupled digital twin model and outputting a three-dimensional physical field state map of the failure area are as follows: Based on the material and structural parameters of the target solid-state battery, synchrotron radiation X-ray tomography is used to reconstruct the three-dimensional voxel matrix inside the electrode with nanometer resolution and obtain the three-dimensional microstructure of the electrode; Based on the three-dimensional voxel matrix, the three-dimensional microstructure of the electrode is discretized into regular grid units. Material parameters and structural parameters are injected into each grid unit to establish the electrochemical-thermal-mechanical coupling equation; Based on the electrochemical-thermomechanical coupling equations, a multi-physics field coupling digital twin model is constructed to dynamically reflect the ion concentration gradient distribution, stress field distribution, and temperature field distribution; Based on the three-dimensional microstructure of the electrode provided by the three-dimensional voxel matrix, the digital twin model coupled with multiple physical fields dynamically outputs the three-dimensional physical field state map of ion concentration gradient distribution, stress distribution field and temperature field distribution; Through the regional anomaly frequency in historical failure cases, the failure area positions of the lithium dendrite growth zone and the interface peeling zone are dynamically identified in the three-dimensional physical field state map, forming a three-dimensional physical field state map of the failure area.

4. The solid-state battery performance testing method based on data analysis according to claim 1, characterized in that: The strain signal and temperature field distribution of the failure area are collected. When abnormal mutations are detected, multi-level warnings are triggered and physical failure evidence is obtained. The steps are as follows: Based on the three-dimensional physical field state map of the failure area, distributed optical fiber strain sensors and non-contact infrared thermal imagers are installed on the surface of the dynamically identified lithium dendrite growth area and interface peeling area to collect local strain signals and surface temperature field distribution in real time; The fiber optic strain sensor continuously outputs strain time series data. When a sudden increase in local strain is detected, it is determined to be the critical point of mechanical failure. The non-contact infrared thermal imager outputs a temperature field distribution map. When the local temperature suddenly changes abnormally, it is determined to be a precursor to thermal runaway. When any of the mechanical failure critical point and thermal runaway precursors occur, multi-level warnings are immediately triggered and a warning record list is obtained; According to the multi-level early warning and safety control, when the early warning is triggered, the physical failure evidence of the lithium dendrite growth area and the interface peeling area is collected through field emission scanning electron microscopy and X-ray photoelectron spectroscopy respectively.

5. The solid-state battery performance testing method based on data analysis according to claim 1, characterized in that: The steps of matching physical failure evidence with historical database, extracting optimal test parameter combination through meta-learning framework, dynamically adjusting charging and discharging strategy and monitoring frequency, executing the adjusted test process, and generating real-time data stream are as follows: The historical database includes crack morphology characteristics, monitoring frequency values, decomposition product components, charge and discharge rate values, pressure loading values and unit cycle failure event frequency; The feature vectorization method is used to convert physical failure evidence and historical database into multidimensional vectors and historical database vectors. The matching degree between the multidimensional vectors and the historical database vectors is calculated by cosine similarity to screen the failure case subset. Based on a subset of failure cases, a radial basis kernel covariance function is used to construct a covariance matrix. The objective function is defined as the negative log marginal likelihood. The cycle life and polarization voltage of the solid-state battery are used as constraints to construct a Gaussian process regression surrogate model. The UE learning framework is used to initialize the search space for charge / discharge rate and pressure loading values. The Gaussian regression proxy model is then used to predict the probability distribution of the charge / discharge rate and pressure loading values. The expected improvement criterion is then used to select the optimal solution that satisfies the constraints and obtain the optimal test parameter combination. According to the predicted charge and discharge rate value in the optimal test parameter combination, the current curve shape of the constant current charge and discharge test is adjusted, and the monitoring frequency value is adjusted according to the pressure loading value to form a test parameter log after adjustment; After the segmented constant current mode and monitoring frequency adjustment are completed, the constant current charge and discharge test is performed through the battery testing method, combined with the DC internal resistance method to calculate the internal resistance and the infrared thermal imager to monitor the temperature, generating a real-time data stream.

6. The solid-state battery performance testing method based on data analysis according to claim 1, characterized in that: The back propagation algorithm is used to update the parameters of the multi-physics field coupled digital twin model and output the revised multi-physics field coupled digital twin model. The specific steps are as follows: The real-time data stream is input into the Levenberg-Marquardt optimizer. By minimizing the residual between the actual observation data and the theoretical prediction value of the multi-physics field coupled digital twin model, a back propagation algorithm is implemented to correct the solid phase diffusion coefficient and interface reaction term in the multi-physics field coupled digital twin model. Combined with the actual cycle capacity attenuation data in the real-time data stream, the degradation equation of the multi-physics field coupled digital twin model is reversely fitted to calibrate the nonlinear cumulative effect of the active material loss rate during the cycle process, and output the corrected multi-physics field coupled digital twin model.

7. The solid-state battery performance testing method based on data analysis according to claim 1, characterized in that: The real-time analysis of the solid-state battery timing sequence, obtaining the failure probability curve, and generating a test report are as follows: Dynamically simulate the charge and discharge cycle process based on the modified multi-physics field coupled digital twin model, analyze the real-time generated time series of voltage, current and temperature, and extract the voltage polarization curve shape and capacity decay rate value; Based on the unit cycle failure event frequency in the historical database, the adjacent difference values of the time series are calculated in real time as the decay rate slope to form the input data set for the Granger causality test; Based on the voltage polarization curve morphology and capacity decay rate value, Granger causality test is used to correlate the capacity decay rate slope with the failure event frequency, generate the cycle number-failure probability curve, and mark the critical threshold; Based on the cycle number-failure probability curve, the 3D physical field state map of the failure area, the multi-level warning record and the adjusted test parameter log, the MCP protocol is used to intelligently align the standardized template with multi-source data to generate a test report.

8. A solid-state battery performance testing system based on data analysis, based on the solid-state battery performance testing method based on data analysis according to any one of claims 1 to 7, characterized in that: include: Failure module, early warning module, test module, correction module and analysis module, The failure module establishes a multi-physics field coupled digital twin model based on the material and structural parameters of the target solid-state battery, and outputs a three-dimensional physical field state map of the failure area; The early warning module collects strain signals and temperature field distribution in the failure area based on the three-dimensional physical field state map of the failure area. When abnormal mutations are detected, multi-level early warnings are triggered and physical failure evidence is obtained; The test module matches physical failure evidence with a historical database, extracts the optimal test parameter combination through a meta-learning framework, dynamically adjusts the charge and discharge strategy and monitoring frequency, executes the adjusted test process, and generates a real-time data stream; The correction module updates the parameters of the multi-physics field coupled digital twin model through the back-propagation algorithm based on the real-time data stream and outputs the corrected multi-physics field coupled digital twin model; The analysis module, based on the revised multi-physics field coupled digital twin model, analyzes the timing sequence of the solid-state battery in real time, obtains the failure probability curve, and generates a test report.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the solid-state battery performance testing method based on data analysis according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the solid-state battery performance testing method based on data analysis according to any one of claims 1 to 7 are implemented.

Images