Modular lithium battery pack non-destructive testing and life prediction system for fire storage joint debugging

Abstract

The application relates to the technical field of lithium battery detection, and discloses a modular lithium battery pack nondestructive detection and life prediction system for fire storage joint debugging. The system comprises a multi-modal data fusion acquisition module, which synchronously captures battery pack temperature fields, stress waves and partial discharge signals through a distributed optical fiber sensing array and an electromagnetic acoustic emission sensor; an electrochemical state reconstruction module adopts a polarization voltage decomposition algorithm to separate three types of polarization components and reconstruct three-dimensional distribution of ion migration paths; a life prediction core engine utilizes a degradation trajectory matching algorithm to establish a capacity attenuation and multi-physical field parameter mapping and generate a prediction vector containing a cycle number and a calendar aging factor; and an abnormality propagation tracing module combines a coupled field simulation to deduce an abnormality propagation path. The system realizes nondestructive detection and accurate life prediction, and facilitates management personnel to quickly master the health distribution condition of the whole lithium battery pack and the specific state of each unit.

Description

Technical Field

[0001] This invention relates to the field of lithium battery testing technology, specifically to a non-destructive testing and life prediction system for modular lithium battery packs designed for integrated thermal and energy storage systems. Background Technology

[0002] Against the backdrop of the rapid development of the new energy industry, lithium battery packs are key equipment for realizing the consumption of new energy. Their operational safety and reliability directly affect the efficient operation of energy storage applications. Under long-term charge and discharge cycles and complex operating conditions, lithium battery packs may experience problems such as capacity decay, abnormal internal polarization, and local thermal runaway. If these hidden dangers cannot be detected and predicted in time, they may cause safety accidents and serious economic losses.

[0003] Currently, lithium battery pack testing technology has many limitations. In terms of data acquisition, traditional testing methods mostly use single sensing methods, which can only capture single physical quantities such as temperature or voltage, making it difficult to comprehensively reflect the coupling effects of multiple physical fields inside the battery. For example, using a temperature sensor alone cannot simultaneously acquire the propagation characteristics of stress waves and partial discharge signals inside the battery, resulting in a one-sided judgment of the battery status; while distributed sensing layouts are prone to data acquisition blind spots, making it impossible to achieve full-area coverage monitoring.

[0004] In the field of electrochemical state analysis, existing technologies lack sufficient precision in separating polarization components, making it difficult to accurately distinguish the contribution ratios of ohmic polarization, concentration polarization, and electrochemical polarization. Consequently, they cannot accurately reconstruct the internal ion migration pathways of the battery. This results in an insufficient understanding of the internal reaction mechanisms of the battery, failing to provide reliable basic data for subsequent lifetime assessment. Furthermore, traditional lifetime prediction methods often rely on a single cycle aging factor, ignoring the impact of calendar aging on battery life, and fail to establish an effective correlation between capacity decay and multiple physical field parameters, leading to significant prediction biases and making it difficult to meet the demand for accurate battery lifetime prediction.

[0005] In terms of anomaly tracing, existing technologies often only locate the individual battery cell where the anomaly occurs, failing to deduce the propagation path of the anomaly between multiple cells. Due to the lack of combined analysis of the battery pack's spatial topology and thermo-mechanical-electrical coupling field characteristics, it is difficult to quickly determine the potential impact range when a local anomaly occurs, posing significant challenges to fault diagnosis and maintenance. Furthermore, current health status displays are mostly numerical representations of single parameters, lacking intuitive hierarchical diagrams, making it difficult for managers to quickly grasp the health distribution of the entire lithium battery pack and hindering the timely development of targeted maintenance strategies. These technical bottlenecks severely restrict the safe and stable operation of lithium battery packs, necessitating an integrated, high-precision non-destructive testing and life prediction system to address these issues. Summary of the Invention

[0006] The purpose of this invention is to provide a non-destructive testing and life prediction system for modular lithium battery packs for integrated thermal and energy storage, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, this invention provides a non-destructive testing and life prediction system for modular lithium battery packs designed for integrated thermal and energy storage systems. The system includes:

[0008] The multimodal data fusion acquisition module uses a distributed fiber optic sensing array and an electromagnetic acoustic emission sensor to simultaneously capture the surface temperature field distribution, internal stress wave propagation characteristics, and partial discharge signals of the lithium battery pack.

[0009] The electrochemical state reconstruction module, based on the heterogeneous sensing data output by the multimodal data fusion acquisition module, separates the ohmic polarization, concentration polarization and electrochemical polarization components through the polarization voltage decomposition algorithm, and reconstructs the three-dimensional distribution of ion migration paths inside the battery.

[0010] The core engine for lifetime prediction uses the polarization components and ion migration paths output by the electrochemical state reconstruction module to establish a mapping relationship between battery capacity decay and multi-physics parameters using a degradation trajectory matching algorithm, generating a lifetime prediction vector that includes cycle number and calendar aging factor.

[0011] The anomaly propagation and source tracing module maps the anomaly prediction nodes output by the lifetime prediction core engine to the battery pack spatial topology, and combines the thermal-mechanical-electric coupling field simulation to deduce the propagation path of anomalies among multiple battery cells.

[0012] The health status visualization module integrates lifespan prediction vectors and anomaly propagation path data to generate a hierarchical health status map of the battery pack.

[0013] Preferably, the multimodal data fusion acquisition module specifically includes:

[0014] The distributed fiber optic temperature sensing unit arranges a spiral-wound fiber optic grid along the surface of the battery module and obtains the temperature gradient distribution through Brillouin dispersion radio frequency shift analysis technology.

[0015] The electromagnetic acoustic emission detection unit uses a wideband piezoelectric transducer array to capture stress wave signals during battery charging and discharging, and uses the wavelet packet energy entropy algorithm to extract the frequency band features of the stress wave.

[0016] The partial discharge monitoring unit collects internal discharge pulse signals of the battery through an ultra-high frequency sensor and identifies the discharge type by combining the pulse phase resolution spectrum.

[0017] The data synchronization and alignment unit performs time stamp calibration on temperature field data, stress wave characteristics and discharge pulse signals, and establishes a timestamp alignment mechanism for cross-modal data.

[0018] Preferably, the electrochemical state reconstruction module specifically includes:

[0019] The dynamic impedance calculation unit performs Hilbert transform on the partial discharge signal output by the multi-modal data fusion acquisition module and extracts the discharge pulse envelope as the excitation response input.

[0020] The polarization component separation unit uses a recursive least squares algorithm to fit the dynamic impedance spectrum curve and decompose it to obtain the ohmic polarization resistance, electrochemical polarization capacitance and concentration polarization time constant.

[0021] The ion migration modeling unit constructs an equivalent circuit model of the porous electrode based on the parameters output by the polarization component separation unit, and calculates the diffusion flux density of lithium ions in the positive and negative electrode materials using the finite element method.

[0022] Preferably, the lifespan prediction core engine specifically includes:

[0023] The degradation feature extraction unit performs principal component analysis on the ion diffusion flux density output by the electrochemical state reconstruction module and selects the top three principal components with the highest correlation to capacity decay as feature vectors.

[0024] The trajectory matching database stores historical degradation trajectory data of lithium batteries with different material systems under various working conditions. Each trajectory includes capacity decay rate, polarization parameter change trend and temperature stress coefficient.

[0025] The real-time matching calculation unit performs dynamic time-normalization comparison between the feature vector of the current battery pack and the historical data in the trajectory matching database, and outputs the five closest reference degradation trajectories and their weight coefficients.

[0026] Preferably, the lifespan prediction core engine further includes:

[0027] The remaining lifetime calculation unit uses a weighted average algorithm to fuse the prediction results of each trajectory based on the reference degradation trajectory output by the real-time matching calculation unit, and generates a lifetime prediction range that includes the median capacity retention rate and upper and lower boundaries.

[0028] The calendar aging correction unit calculates the impact of temperature acceleration factor on the non-cycle aging portion of the battery based on the Arrhenius equation, and dynamically adjusts the boundary values ​​of the lifetime prediction range.

[0029] Preferably, the anomaly propagation tracing module specifically includes:

[0030] The battery pack topology modeling unit constructs a three-dimensional mesh model based on the physical connection method of the modular lithium battery pack, and marks the thermal conductivity coefficient and current path impedance between each battery cell.

[0031] The multi-field coupled simulation unit uses the abnormal nodes output by the lifetime prediction core engine as the initial excitation source and solves the coupled differential equations of the thermal field and stress field through an implicit algorithm in alternating directions.

[0032] The propagation path marker unit records the sequence of grid cells where abnormal parameters exceed the threshold during the simulation process, forming a directed propagation link graph of abnormal diffusion.

[0033] Preferably, the anomaly propagation tracing module further includes:

[0034] The critical state early warning unit generates an early warning signal containing the affected battery number when the abnormal diffusion range output by the propagation path marking unit touches the battery pack boundary unit.

[0035] The failure mode classification unit matches a library of typical failure modes based on the spatial distribution characteristics of abnormal propagation paths, and outputs a short-circuit risk level or thermal runaway probability index.

[0036] Preferably, the health status visualization module specifically includes:

[0037] The 3D rendering engine converts the ion migration path data output by the electrochemical state reconstruction module into a color gradient encoded 3D streamline map.

[0038] The lifespan degradation heatmap generates color temperature gradient effects for each cell in the battery pack based on the predicted range values ​​output by the lifespan prediction core engine.

[0039] The anomaly diffusion animation unit generates a dynamic diffusion process simulation image based on the directed propagation link diagram output by the anomaly propagation tracing module.

[0040] Preferably, the health status visualization module further includes:

[0041] The interactive diagnostic unit responds to externally inputted battery cell selection commands and retrieves the historical polarization parameter curves and current ion migration rate data for the corresponding cell.

[0042] The comparative analysis view displays the differences in health status indicators of similar battery modules at the same number of cycles.

[0043] Preferably, the system further includes:

[0044] The thermal power generation and storage joint commissioning interface module receives the dispatching instructions from the thermal power generating unit and calculates the optimal charging and discharging power range of the battery pack based on the life prediction vector.

[0045] The dynamic parameter configuration unit adjusts the sampling frequency of the multimodal data fusion acquisition module and the calculation accuracy level of the electrochemical state reconstruction module in real time.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] Through the innovative design of the multimodal data fusion acquisition module, and by adopting a collaborative approach between a distributed fiber optic sensing array and an electromagnetic acoustic emission sensor, the synchronous acquisition of the surface temperature field distribution, internal stress wave propagation characteristics, and partial discharge signals of the lithium battery pack is achieved. This breaks through the limitations of traditional single sensing methods, enabling comprehensive acquisition of multi-dimensional physical information during battery operation and fully reflecting the coupling state of multiple physical fields inside the battery. This provides rich and comprehensive basic data for subsequent state analysis and life prediction.

[0048] The polarization voltage decomposition algorithm employed in the electrochemical state reconstruction module can accurately separate ohmic polarization, concentration polarization, and electrochemical polarization components, effectively solving the problem of ambiguous differentiation of polarization components in traditional technologies. Based on this separation result, the reconstructed three-dimensional distribution of ion migration paths within the battery clearly presents the dynamic laws of ion migration, providing an intuitive analytical basis for a deeper understanding of the internal electrochemical reaction mechanism of the battery, and making the assessment of the battery state more closely aligned with its actual operating conditions.

[0049] The core lifespan prediction engine combines electrochemical state reconstruction results with a degradation trajectory matching algorithm to establish a mapping relationship between battery capacity decay and multi-physics parameters. It not only considers the impact of cycle count on battery lifespan but also incorporates calendar aging factors, constructing a more comprehensive lifespan assessment system. The generated lifespan prediction vector accurately reflects the remaining battery lifespan, providing a scientific reference for lithium battery pack replacement and maintenance planning. This helps to rationally allocate operation and maintenance resources and reduce system downtime losses caused by sudden battery failures.

[0050] The anomaly propagation and tracing module combines anomaly prediction nodes with the spatial topology of the battery pack. Through thermo-mechanical-electrical coupled field simulation technology, it deduces the propagation path of anomalies among multiple battery cells, enabling rapid location of the anomaly source and its potential impact range. This departs from the traditional approach of only being able to identify anomaly cells in isolation. This function allows maintenance personnel to take timely and targeted preventative measures, interrupting the anomaly propagation chain, reducing the risk of fault escalation, and ensuring the overall operational safety of the lithium battery pack.

[0051] The health status visualization module integrates lifetime prediction vectors and anomaly propagation path data to generate a hierarchical health status map. This presents complex detection and prediction data in an intuitive form, allowing managers to quickly grasp the health distribution of the entire lithium battery pack and the specific status of each unit. This visualization method simplifies the data interpretation process, improves battery management efficiency, and enables more timely and accurate decision-making, further ensuring the stable and efficient operation of the lithium battery pack. Attached Figure Description

[0052] Figure 1This is a schematic diagram of the working principle of the modular lithium battery pack non-destructive testing and life prediction system for integrated thermal and energy storage as described in this invention.

[0053] Figure 2 A flowchart illustrating the operation of the multimodal data fusion and acquisition module;

[0054] Figure 3 A flowchart illustrating the operation of the lifespan prediction core engine. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Please see Figure 1 This invention provides a modular lithium-ion battery pack non-destructive testing and lifetime prediction system for integrated thermal and energy storage systems. The system includes: a multimodal data fusion acquisition module, an electrochemical state reconstruction module, a lifetime prediction core engine, an anomaly propagation tracing module, and a health status visualization module. The multimodal data fusion acquisition module is deployed on the surface and inside the battery pack, employing a distributed fiber optic sensing array and electromagnetic acoustic emission sensors for synchronous measurement, capturing temperature field distribution, stress wave propagation characteristics, and partial discharge signals. These heterogeneous data are integrated into a unified dataset through a timestamp alignment mechanism. The electrochemical state reconstruction module receives the output from the multimodal data fusion acquisition module, uses a polarization voltage decomposition algorithm to analyze ohmic polarization, concentration polarization, and electrochemical polarization components, and reconstructs the internal electrochemical state of the battery based on the three-dimensional distribution of ion migration paths. The lifetime prediction core engine utilizes the polarization parameters and ion migration data provided by the electrochemical state reconstruction module, establishes a mapping relationship between capacity decay and multi-physics parameters through a degradation trajectory matching algorithm, and generates a lifetime prediction vector including cycle number and calendar aging factor. The anomaly propagation and tracing module maps anomaly nodes identified by the lifetime prediction core engine to the battery pack's spatial topology, and uses thermo-mechanical-electrical coupling field simulation to deduce the propagation path of anomalies among multiple battery cells. The health status visualization module integrates lifetime prediction vectors and anomaly propagation path data to generate a hierarchical health status map, supporting visual monitoring.

[0057] Example 1: See Figure 2The distributed fiber optic temperature sensing unit of the multimodal data fusion acquisition module is arranged along the surface of the battery module. A spiral-wound fiber optic grid structure covers the entire outer surface of the battery module. The fiber optic grid is composed of multi-core optical fibers to improve spatial resolution. The grid point spacing of each fiber is set to 5 mm to achieve dense temperature measurement. The Brillouin scattering frequency shift analysis technology obtains the temperature gradient distribution by injecting continuous wave laser into the optical fiber and analyzing the frequency shift of the backscattered light. The laser interferometer detects the frequency shift of the scattered light. The frequency shift is linearly related to the temperature change. The algorithm calculates the absolute temperature value of each grid point based on the frequency shift difference and generates a two-dimensional temperature field matrix. The electromagnetic acoustic emission detection unit mounts a wideband piezoelectric transducer array at the center of each of the six faces of the battery casing. The transducer array consists of sixteen piezoelectric ceramic plates arranged in a 4x4 matrix, with each plate's resonant frequency covering a bandwidth from 20kHz to 1MHz. Stress wave signals generated during battery charging and discharging are converted into electrical signals by the transducers. These signals are then amplified by a preamplifier before entering the data acquisition card. A wavelet packet energy entropy algorithm performs six-level wavelet decomposition on the acquired stress wave data, extracting energy distribution characteristics across eight frequency bands. The energy entropy value is calculated based on the variance of the wavelet coefficients in each frequency band and is used to identify abnormal patterns of internal battery structural changes. The partial discharge monitoring unit embeds ultra-high frequency sensors into the gaps between battery cells. These sensors employ a microstrip patch antenna design, with the peak sensitivity adjusted to around 500MHz to capture discharge pulses. Pulse phase-resolved spectral analysis groups the acquired pulse sequences according to the grid phase, generating a two-dimensional distribution map of the pulse amplitude relative to the phase. A density clustering algorithm is then used to distinguish the characteristic spectra of three discharge types: corona discharge, surface discharge, and internal discharge. The data synchronization and alignment unit uses a GPS-disciplined crystal oscillator as a clock source to provide a unified time reference for all sensor data streams, with a timestamp accuracy of 1 microsecond. Temperature field data, stress wave characteristics and discharge pulse signals with different sampling rates are resampled to a unified frequency of 100kHz through a cubic spline interpolation algorithm, establishing a timestamp alignment mechanism for cross-modal data and forming a multimodal dataset that is fully synchronized in the time dimension.

[0058] The dynamic impedance calculation unit of the electrochemical state reconstruction module receives the partial discharge pulse signal output by the multimodal data fusion acquisition module. Hilbert transform is applied to extract the envelope waveform from the pulse sequence. The transformation process is achieved by convolving the pulse signal with the Hilbert kernel function. The resulting envelope is used as the excitation response and input to the impedance calculation model. The model converts the time-domain envelope into a frequency-domain impedance spectrum using a fast Fourier transform. The real and imaginary parts of the impedance spectrum are decomposed into resistance and reactance components. The polarization component separation unit uses a recursive least squares algorithm to fit the dynamic impedance spectrum curve. The algorithm is initialized with a forgetting factor of 0.95 to balance the weights of historical and new data. The recursive process updates the equivalent circuit model parameters in real time based on the impedance spectrum data points. The equivalent circuit model includes a topology of ohmic polarization resistance in series with electrochemical polarization capacitance and then in parallel with concentration polarization impedance. The decomposition results output three key parameters: ohmic polarization resistance value, electrochemical polarization capacitance value, and concentration polarization time constant. The ion migration modeling unit constructs an equivalent circuit model of the porous electrode based on the parameters output by the polarization component separation unit. The model parameters include the porosity of the positive electrode material, the specific surface area of ​​the negative electrode material, and the ionic conductivity of the electrolyte. The finite element method discretizes the battery geometry into 100,000 tetrahedral mesh elements. Each mesh node solves the Nernst-Planck equation to calculate the diffusion flux driven by the lithium ion concentration gradient. The diffusion flux density is output as a three-dimensional vector field data format, where the vector direction represents the ion migration path and the modulus represents the migration rate.

[0059] The fiber optic grid of the distributed fiber optic temperature sensing unit is protected by a high-temperature resistant polyimide coating with a thickness of 50 micrometers to resist potential localized overheating on the battery surface. The pitch of the helical winding is dynamically adjusted according to the battery module size to ensure that each battery cell surface is covered by at least three fiber optic grid points. The laser source for the Brillouin frequency shift analysis technology uses a distributed feedback laser with a stable output power of 10 dBm. The backscattered light is separated by a circulator and then enters an interferometer to detect the frequency shift. The temperature analysis algorithm is based on the linear relationship between the Brillouin frequency shift and temperature, and the coefficient is determined to be 1.2 MHz / ℃ through standard temperature calibration experiments. The piezoelectric transducer array of the electromagnetic acoustic emission detection unit is encapsulated in epoxy resin with a thickness of 2 mm to achieve acoustic impedance matching. The decomposition basis function of the wavelet packet energy entropy algorithm is the Db4 wavelet, and the entropy value is calculated using the Shannon entropy formula. When the energy entropy value of a certain frequency band exceeds a threshold, an anomaly recording is triggered. The UHF sensor of the partial discharge monitoring unit is connected to the signal conditioning circuit via a coaxial cable. The cable shielding layer adopts a double-layer braided structure to suppress electromagnetic interference. The phase window width of the pulse phase resolution spectrum analysis is set to 10 degrees, and pulse data of 360 phase points are collected in each cycle. The clustering algorithm uses the DBSCAN method to identify the discharge type.

[0060] The data synchronization and alignment unit outputs a 10MHz reference clock from its GPS-disciplined crystal oscillator, which synchronizes the sampling clocks of all data acquisition cards via a clock distributor. The interpolation algorithm employs piecewise cubic Hermitian interpolation to ensure data smoothness after resampling. Time-aligned data packets are added with timestamp indices and stored in a circular buffer. The Hilbert transform of the dynamic impedance calculation unit is calculated using a finite-length unit impulse response filter implemented by a digital signal processor. The filter order is set to 128 to balance computational accuracy and real-time performance. The Fourier transform uses a radix-2 fast Fourier transform algorithm with a fixed number of 1024 transform points. The recursive least squares algorithm of the polarization component separation unit is executed in a pipelined manner within the embedded system, with the parameter update cycle synchronized with the battery charge / discharge cycle. The concentration polarization impedance in the equivalent circuit model is modeled using Warburg elements, and the element parameters are obtained by fitting the impedance spectrum arc characteristics using a nonlinear least squares method. The finite element mesh generation of the ion migration modeling unit adopts the advancing front method, and the mesh size is refined to 0.1 mm in the electrode active material region. The Nernst-Planck equation is solved using the Galerkin finite element discretization scheme, and the diffusion coefficient is dynamically adjusted as a field variable with temperature. The temperature data comes from the real-time input of the multimodal data fusion acquisition module.

[0061] The distributed fiber optic temperature sensing unit of the multimodal data fusion acquisition module shares a communication bus with the battery management system, and temperature data is transmitted via the CAN bus protocol. The signal acquisition dynamic range of the electromagnetic acoustic emission detection unit is set to ±10V to cover stress wave signal amplitude variations. The sensitivity calibration of the ultra-high frequency sensor in the partial discharge monitoring unit adopts the standard pulse source injection method, with a calibration cycle of once every 24 hours. The dynamic impedance calculation unit of the electrochemical state reconstruction module sets impedance calculation trigger conditions, initiating the calculation process when the amplitude of the local discharge pulse exceeds a set threshold. The recursive least squares algorithm of the polarization component separation unit sets a convergence criterion, stopping iteration when the parameter update amount is less than 0.05%. The finite element calculation of the ion migration modeling unit adopts a multi-core parallel processing architecture, with the calculation task distributed across eight processor cores to simultaneously solve the mesh node equations. The data synchronization and alignment unit of the multimodal data fusion acquisition module has a data integrity verification function, using cyclic redundancy check codes to verify errors during data transmission. The output data format of the ion migration modeling unit of the electrochemical state reconstruction module conforms to the OpenFOAM field data standard, facilitating direct use by subsequent visualization modules.

[0062] Example 2: See Figure 3The degradation feature extraction unit receives ion diffusion flux density data output by the electrochemical state reconstruction module. The ion diffusion flux density data represents the ion migration rate in different regions inside the battery in the form of a three-dimensional vector field. The principal component analysis algorithm calculates the covariance matrix of the multi-dimensional flux density sequence. After the matrix eigenvalues ​​are decomposed, the eigenvalues ​​are arranged in descending order. The eigenvectors corresponding to the three largest eigenvalues ​​are selected as principal components. The contribution rate of the principal components is calculated by the proportion of each eigenvalue to the sum of the eigenvalues. The eigenvectors are normalized so that each component is within the range of zero to one. The selection result is output to the subsequent unit as the eigenvector with the highest correlation to capacity decay. The trajectory matching database adopts a distributed architecture to store historical degradation trajectory data of lithium batteries with different material systems under various operating conditions. The database is partitioned according to the type of battery cathode material into lithium iron phosphate partition, ternary material partition and lithium cobalt oxide partition. Each partition stores the cycle test records of the corresponding material battery under constant current charge and discharge, variable current pulse charge and discharge and different ambient temperatures. Each trajectory data includes a numerical sequence of capacity decay rate, polarization parameter change trend curve and temperature stress coefficient matrix. The database index establishes a multi-dimensional tag system based on material type, upper limit of cycle number and temperature range. Data compression adopts lossless compression algorithm to reduce storage space occupation.

[0063] The real-time matching unit calls the feature vector output by the degradation feature extraction unit and performs dynamic time warping comparison with historical data in the trajectory matching database. The dynamic time warping algorithm constructs a cost matrix between the current feature vector and the historical trajectory feature sequence. The matrix element calculation uses Euclidean distance to measure the difference between corresponding points. The algorithm finds a path with the minimum cumulative cost that runs through the cost matrix. The path distance value is used as the similarity evaluation criterion. The comparison process outputs the five reference degradation trajectories that are closest to the current battery pack state and their weight coefficients. The weight coefficients are normalized and assigned according to the reciprocal of the path distance value. The remaining lifetime calculation unit receives the five reference degradation trajectories and their weight coefficients output by the real-time matching unit. The weighted average algorithm fuses the prediction results of each trajectory. The weight coefficient is multiplied by the predicted remaining cycle number of the corresponding trajectory. The sum of the products is divided by the sum of the weight coefficients to obtain the median predicted capacity retention rate. The upper and lower boundaries of the lifetime prediction interval are determined by calculating the percentiles of the predicted values ​​of each trajectory. The 10th percentile is used as the lower boundary, and the 90th percentile is used as the upper boundary.

[0064] The calendar aging correction unit calculates the impact of the temperature acceleration factor on the non-cycle aging portion of the battery based on the Arrhenius equation. The input parameters for the Arrhenius equation include the battery activation energy constant and the real-time ambient temperature. The activation energy constant is selected from a preset parameter table according to the battery material type. The product of the temperature acceleration factor calculation results is applied to the non-cycle aging baseline rate, which is derived from accelerated aging experimental data of the battery material. The corrected value dynamically adjusts the boundary values ​​of the lifetime prediction interval, and the adjustment cycle is synchronized with the battery management system data update cycle. During the execution of the principal component analysis algorithm in the degradation feature extraction unit, the ion diffusion flux density data is first standardized to eliminate dimensional influences. The standardization formula uses the Z-score method to convert the data into a distribution with a mean of zero and a standard deviation of one. The covariance matrix eigenvalue decomposition is implemented using the Jacobi rotation algorithm, with the iterative convergence threshold set to 10 to the power of negative six. The historical degradation trajectory data in the trajectory matching database is collected from over a thousand battery cycle test experiments, covering temperature ranges from -20°C to 60°C. The data storage format adopts the fifth-generation hierarchical data format, with each data file containing complete metadata description information, including fields such as battery model, test start time, and charge / discharge protocol version. The dynamic time warping algorithm of the real-time matching computing unit employs a dynamic programming optimization strategy, keeping the algorithm's time complexity within O(n*m) to meet real-time requirements. The cost matrix filling process utilizes parallel computing technology, dividing the matrix into multiple sub-blocks to simultaneously calculate distance values.

[0065] The weighted average algorithm of the remaining lifetime calculation unit introduces a weight decay mechanism, where the weight coefficients of earlier historical data gradually decrease over time, and the decay function adopts an exponential decay model. Percentile calculation uses nearest neighbor interpolation to determine boundary values, and the interpolation parameters are dynamically adjusted according to the data distribution density. The temperature acceleration factor calculation module of the calendar aging correction unit establishes a data interface with the multimodal data fusion acquisition module, and the real-time ambient temperature value is taken from the average temperature value output by the distributed fiber optic temperature sensing unit. The non-cyclic aging baseline correction model considers the temperature accumulation effect and calculates the correction coefficient by integrating the continuous high-temperature exposure time. The output of the lifetime prediction core engine is in a structured data format, including the median prediction of the remaining cycle count of the battery pack, the upper and lower boundary values ​​of the prediction interval, and confidence level indicators. The data packets are transmitted to the anomaly propagation tracing module for further processing via a secure encryption protocol.

[0066] The trajectory matching database establishes a data update mechanism, with newly added experimental data added to the database storage scope every quarter. A data deduplication algorithm based on feature vector similarity avoids duplicate records. The database query interface supports multi-condition queries, including battery material, cycle number range, and temperature range. The real-time matching calculation unit sets matching quality evaluation indicators. When the similarity between the five reference degradation trajectories and the current state is all below a threshold, an early warning signal is triggered, notifying the system administrator for manual intervention. The remaining lifetime calculation unit incorporates uncertainty propagation theory into its lifetime prediction interval calculation, considering the impact of input parameter errors on the output interval width. The uncertainty propagation model uses the Monte Carlo method to simulate the changes in the distribution of predicted values ​​caused by parameter fluctuations. The environmental temperature data preprocessing module of the calendar aging correction unit includes an outlier filtering function, using a sliding window detection method to identify and remove abnormal temperature sensor readings. The correction calculation process uses an incremental update method, recalculating only the portion that has changed since the last update.

[0067] A bidirectional data verification mechanism is established between the degradation feature extraction unit and the electrochemical state reconstruction module of the lifespan prediction core engine. After feature vector generation, the consistency with the original ion migration path data is verified in reverse. Access control is implemented for the trajectory matching database, with different user roles having different data query and modification permissions; operation logs record all database access behaviors for security auditing. The dynamic time warping algorithm of the real-time matching calculation unit is optimized for battery data characteristics, introducing local path constraints to limit the curvature of the matching path to avoid excessive distortion. The weighted average algorithm of the remaining lifespan calculation unit supports manual adjustment of weight coefficients, allowing expert users to fine-tune the weight allocation of different trajectories based on experience. The temperature acceleration factor calculation of the calendar aging correction unit considers the non-uniform temperature distribution within the battery pack, introducing a temperature field weighted average algorithm to improve correction accuracy. The hardware platform of the lifespan prediction core engine adopts an industrial-grade server architecture, equipped with multiple high-performance processors and large-capacity memory; the software system implements a modular design, with asynchronous communication between units via message queues. The principal component analysis algorithm of the degradation feature extraction unit is deployed on a mathematical computing accelerator card, utilizing parallel computing capabilities to shorten computation time. The trajectory matching database implements a regular backup strategy, with backup data stored in an off-site disaster recovery center to ensure data security. The real-time matching unit's matching result caching mechanism stores the results of the most recent 1,000 queries, directly returning cached data for the same query conditions to improve response speed. The remaining lifetime calculation unit's prediction result visualization module generates a lifetime decay curve, showing the capacity retention rate trend over time and the predicted range. The calendar aging correction unit's temperature data calibration module periodically compares its data with a standard temperature source, with a calibration cycle set to once every thirty days.

[0068] Example 3: The battery pack topology modeling unit constructs a 3D mesh model based on the physical connection method of the modular lithium battery pack. The model is built based on the actual geometric dimensions and connection structure of the battery pack, with each battery cell abstracted as a hexahedral mesh element. Mesh nodes are located at the geometric center of the battery cell, and the node coordinates are obtained through actual position data acquired using a 3D measuring instrument. Mesh edges represent the heat conduction paths and electrical connection paths between battery cells. The thermal conductivity coefficient is assigned based on a database of thermal conductivity parameters for the battery casing material, including the thermal conductivity of the aluminum alloy casing. Thermal conductivity of epoxy resin insulation layer The current path impedance is measured using a DC internal resistance tester to measure the resistance between each connection point. After the measurement data is imported into the model, the impedance parameters of the mesh edges are automatically calibrated. The 3D mesh model employs unstructured mesh generation technology, with the mesh size adaptively refined based on the spacing between battery cells. In the battery cell contact area, the mesh size is reduced to a smaller value. To improve calculation accuracy.

[0069] The multi-field coupled simulation unit uses the anomalous nodes output by the lifetime prediction core engine as the initial excitation source, mapping these anomalous nodes to corresponding positions in the 3D mesh model. The solution of the coupled differential equations of the thermal and stress fields employs an alternating direction implicit algorithm, which decomposes the 3D problem into three alternating 1D problems. The thermal field governing equation is based on Fourier's law of heat conduction and is expressed as follows:

[0070] ,

[0071] in: Indicates the density of battery materials. This indicates the specific heat capacity of the battery material. Indicates the temperature field distribution. Represents a time variable. Represents the thermal conductivity tensor. Represents the vector differential operator. The term represents the heat source. The stress field governing equations are based on linear elasticity theory, considering the influence of thermal expansion. In the coupled solution process, the temperature field distribution is first solved at each time step. The temperature distribution result is then used as the thermal stress input to the stress field equations, and the stress field calculation results are fed back to influence the nonlinear changes in material parameters. The discretization scheme of the alternating direction implicit algorithm uses the central difference approximation, and the time step is automatically adjusted according to the Courant number condition to ensure numerical stability.

[0072] The propagation path marking unit monitors the output data of the multi-field coupled simulation unit in real time and records the sequence of mesh elements whose abnormal parameters exceed the threshold during the simulation process; the abnormal parameter threshold is set to the normal value of the parameter. The threshold benchmark is derived from parameter measurements taken during the initial health state of the battery pack. The monitoring process employs a sliding window detection algorithm with a window size of ten time steps. A grid cell is marked as abnormal when its abnormal parameters continuously exceed the threshold. Sequence data is stored as a directed graph structure, where nodes represent abnormal grid cells, edges represent the direction of abnormal propagation, and edge weights represent propagation intensity. The directed graph structure uses an adjacency list for storage, supporting fast query and traversal operations. The critical state early warning unit continuously monitors the abnormal propagation range output by the propagation path marking unit. An early warning signal is generated when the abnormal propagation path touches the battery pack boundary cell; the boundary cell is defined as the outermost set of grid cells in the battery pack's 3D mesh model. The early warning signal includes a list of affected battery numbers, mapped to the battery pack's physical identifier, and records the timestamp of the abnormality reaching the boundary and the abnormal parameter values. The early warning signal is asynchronously sent to the battery management system via a message queue, using the JSON standard for easy parsing and processing. The failure mode classification unit matches a typical failure mode library (TML) with the spatial distribution characteristics of the anomaly propagation path. The TML contains standard propagation path templates for fault types such as short circuits, internal short circuits, and thermal runaway. The matching algorithm uses graph similarity calculation to determine the maximum common subgraph ratio between the current anomaly propagation directed graph and the template directed graph. Graph similarity calculation is based on adjacency matrix eigenvalue decomposition, extracting spectral features of the graph structure for comparison. The matching result outputs a short circuit risk level or a thermal runaway probability index. The risk level is divided into three levels: low risk, medium risk, and high risk. The probability index uses... arrive The numerical representation of .

[0073] The 3D mesh model of the battery pack topology modeling unit supports dynamic updating, automatically regenerating the mesh when battery pack modules are replaced or reassembled. A mesh quality check algorithm verifies the positive determinant condition of the Jacobian matrix of the mesh cells, avoiding distorted meshes from affecting computational accuracy. The heat source term of the multi-field coupled simulation unit... The calculations consider the combined effects of Joule heating and reaction heat. Joule heating is calculated based on current density distribution, while reaction heat is estimated based on an electrochemical model. The threshold management module of the propagation path marking unit supports adaptive adjustment, dynamically optimizing the threshold size based on the battery pack's operating history. The boundary detection algorithm of the critical state early warning unit considers the influence of the battery pack's packaging structure, treating the heat dissipation shell as an isolation boundary. The typical failure mode library of the failure mode classification unit has an update mechanism; newly added failure cases are included in the template library after expert verification. The parallel computing architecture of the multi-field coupled simulation unit divides the 3D mesh model into multiple sub-regions, each assigned to different processor cores for parallel solving; the data exchange interface handles variable transfer at sub-region boundaries. The directed graph construction algorithm of the propagation path marking unit adopts an incremental update strategy, processing only newly marked anomalous units each time to reduce computational overhead. The early warning signal generation module of the critical state early warning unit includes a multi-level early warning mechanism, sending early warnings of different urgency levels according to the anomaly propagation speed. The graph similarity calculation of the failure mode classification unit introduces fuzzy matching technology to handle local differences between the actual propagation path and the standard template.

[0074] The mesh generation algorithm for the battery pack topology modeling unit supports various battery arrangements, including series, parallel, and hybrid structures. The mesh node numbering follows the right-hand rule to determine the local coordinate system orientation. The material parameter database for the multi-field coupled simulation unit includes temperature-related nonlinear parameters and thermal conductivity. and specific heat capacity The relationship is set as a function of temperature. The anomaly parameter monitoring of the propagation path marking unit covers multiple physical quantities, including temperature, stress, and current fields, with independent thresholds and monitoring strategies set for each. The boundary unit identification algorithm of the critical state early warning unit considers the thermal short-circuit effect of the battery pack mounting bracket, treating the bracket contact points as potential boundary points. The risk assessment model of the failure mode classification unit incorporates a time dimension, considering the impact of the anomaly propagation rate on the risk level.

[0075] The anomaly propagation tracing module establishes a data interface with the health status visualization module, converting propagation path data into a streaming data format supported by the visualization engine. The battery pack topology modeling unit provides mesh model export functionality, supporting standard CAE software formats for easy third-party analysis. The multi-field coupling simulation unit's calculation result caching mechanism stores the most recent ten simulation data sets, allowing direct access to cached results under the same initial conditions. The propagation path marking unit's directed graph data is stored in a graph database, supporting efficient querying of complex topological relationships. The critical state early warning unit saves its early warning history to a log file, which is segmented by time for easy post-event analysis. The failure mode classification unit outputs a structured report based on the matching results, including fault type diagnosis, confidence score, and handling recommendations.

[0076] Example 4: The 3D rendering engine converts the ion migration path data output by the electrochemical state reconstruction module into a color gradient-encoded 3D streamline map. The streamline map generation is based on a particle system tracking algorithm, which deploys virtual particles in 3D space. The particle trajectories follow the vector field direction of the ion migration path. Color mapping uses the HSV color space, with hues ranging from blue to red representing changes in ion migration rate from low to high. Saturation is fixed at 100%, and brightness is inversely proportional to the radius of curvature of the migration path. The 3D streamline map is rendered using ray casting technology. The color value of each pixel is calculated by integrating and accumulating particle density along the viewing direction. The transparency function uses an exponential decay model to highlight high migration rate areas. The rendering result is output as a 1920×1080 resolution frame sequence, maintaining a frame rate of 30 frames per second to meet smooth visualization requirements. The lifespan degradation heatmap generates color temperature gradient effects for each cell in the battery pack based on the predicted interval values ​​output by the lifespan prediction core engine. The predicted interval values ​​include the median capacity retention rate and upper and lower boundary data. The color temperature mapping relationship is established based on the CIE1931 color space; the xchromaticity coordinate values ​​are linearly related to the capacity retention rate, and the ychromaticity coordinate values ​​are quadratic functions related to the prediction uncertainty range. The grid division of the heatmap strictly corresponds to the physical structure of the battery pack, with each battery cell corresponding to a color patch. Bilinear interpolation is used within the patch to achieve smooth color transitions. Gamma correction technology is used during rendering to compensate for the nonlinear response of the display device, with a correction coefficient set to 2.2 to conform to the sRGB standard. The anomaly diffusion animation unit generates a dynamic diffusion process simulation image based on the directed propagation link graph output by the anomaly propagation tracing module. The directed propagation link graph includes the anomaly node sequence and propagation timestamp information. The animation generation algorithm uses keyframe interpolation technology; keyframes correspond to the moment the anomaly touches each new node, and intermediate frames calculate the camera path and node state through spherical linear interpolation. The visual representation of the anomalous diffusion adopts the form of pulse wave propagation. The wavefront is rendered as a semi-transparent sphere, the wavefront thickness is proportional to the anomalous intensity, and the wavefront propagation speed is dynamically calculated based on the time difference between adjacent nodes.

[0077] The interactive diagnostic unit responds to externally inputted battery cell selection commands, supporting both mouse click and touchscreen interaction modes. The cell selection algorithm employs ray-projection collision detection, emitting a detection ray from the viewpoint and finding its intersection with the bounding boxes of the battery cells, returning the cell number of the nearest intersection point. Historical polarization parameter curves are plotted using the Canvas 2D drawing API, with the horizontal axis representing the time series and the vertical axis representing the polarization resistance value. Curve smoothing utilizes cubic spline interpolation. Current ion migration rate data is displayed as a floating digital panel with a grid layout, highlighting important parameters in larger font. The comparative analysis view displays side-by-side health status differences between similar battery modules at the same cycle count. The view layout uses a multi-column, equal-height design, with each column corresponding to the monitoring data of one battery module. The difference index is calculated using a relative deviation algorithm, dividing the difference between the current value and the baseline value by the baseline value to obtain the percentage deviation. The view update mechanism is based on a data-driven architecture, automatically re-rendering all visualization components when new monitoring data arrives.

[0078] The parameter configuration of the health status visualization module is stored in a structured configuration file in YAML format for easy manual reading and editing. The shader program of the 3D rendering engine is written in GLSL, with the vertex shader handling coordinate transformations and the fragment shader handling lighting calculations. The color map table for the lifespan decay heatmap is predefined as a 256-color gradient palette, with palette data sourced from professional color design schemes. The timeline controller of the anomaly diffusion animation unit supports slow-motion playback and frame-by-frame analysis to meet different detail observation needs. The data query interface of the interactive diagnostic unit uses the WebSocket protocol for real-time communication, ensuring low latency in interactive responses. The rendering engine of the comparative analysis view is based on SVG vector graphics technology, and graphic elements support lossless scaling to adapt to different display sizes. The health status visualization module establishes a standardized data interface with the underlying data layer, defined using the Protocol Buffers serialization format. The model data of the 3D rendering engine uses octree spatial indexing to accelerate rendering, and the view frustum pruning algorithm removes invisible mesh cells. The algorithm for generating the lifespan decay heatmap incorporates an edge enhancement filter and uses the Sobel operator to detect battery cell boundaries to enhance visual differentiation. The physical simulation of the anomaly propagation animation unit uses a point mass spring system, transforming the anomaly propagation path into the vibration propagation of a spring network. The historical data retrieval in the interactive diagnostic unit employs a B+ tree index structure, supporting fast queries by time range. The calculation of difference indicators in the comparative analysis view incorporates data standardization to eliminate the influence of different units on the comparison results.

[0079] Referring to Table 1, the user interface of the health status visualization module follows ergonomic design principles, with key control elements placed within the interface's hot zone. The 3D rendering engine's camera control system supports both first-person and third-person perspectives, with smooth transition animations during viewpoint switching. The legend for the lifespan decay heatmap uses dynamic annotation technology, displaying detailed numerical information when the mouse hovers over it. The anomaly propagation animation unit's status recording function saves key snapshots during animation playback, with snapshot data including timestamps and camera parameters. The interactive diagnostic unit's interface layout adopts a responsive design, automatically adapting to display devices with different resolutions. The data export function of the comparative analysis view supports both PNG bitmap and PDF vector formats. The health status visualization module's performance optimization employs a multi-level caching strategy, caching geometric data in video memory and texture data in system memory. The 3D rendering engine's detail level management dynamically adjusts model accuracy based on viewpoint distance, using simplified models for long-distance displays to improve rendering efficiency. The color calculation for the lifespan decay heatmap incorporates hardware acceleration technology, utilizing GPU parallel computing capabilities for real-time rendering. The anomaly propagation animation unit's memory management uses an object pool model, reusing animation objects to reduce garbage collection frequency. The interactive diagnostic unit employs a multi-threaded architecture for data preprocessing, separating user interface interaction from data computation and executing them in different threads. The rendering optimization of the comparative analysis view utilizes the dirty rectangle technique, redrawing only the changed areas to reduce unnecessary redrawing operations.

[0080] Table 1: Color Mapping Specifications for Health Status Visualization Module

[0081]

[0082] The health status visualization module employs a multi-layered try-catch structure for exception handling, and detailed log information is recorded for graphics rendering exceptions. Compatibility testing of the 3D rendering engine covers various graphics card models, with test resolutions ranging from 1366×768 to 3840×2160. The color accessibility design of the lifespan decay heatmap considers the needs of colorblind users, offering multiple color themes to choose from. The performance monitoring of the anomaly diffusion animation unit tracks frame rate changes in real time, automatically reducing rendering quality when the frame rate falls below a threshold. User operation records in the interactive diagnostic unit are saved to the operation log, which includes the time, user operation type, and operation parameters. The print output function of the comparative analysis view supports page layout adjustments, and header and footer information can be customized. The installation and deployment of the health status visualization module utilizes containerization technology, with the Docker image containing all runtime dependencies. The debug mode of the 3D rendering engine supports wireframe display and vertex annotation, facilitating developers to inspect model data. The data verification mechanism of the lifespan decay heatmap checks the input data range, triggering data quality control alarms for abnormal values. The scripted control of the anomaly diffusion animation unit supports automated testing, with test scripts simulating user interaction operations. The interactive diagnostic unit's help system integrates context-sensitive help, displaying relevant help information based on the current operation status. The data tracing function in the comparative analysis view records the data source for each indicator; clicking on an indicator displays the original data query statement.

[0083] Example 5: The thermal power and energy storage joint commissioning interface module establishes a data connection with the thermal power generating unit control system through the standard IEC-61850 communication protocol. The communication server is configured with dual network card binding technology to achieve network redundancy, and the data acquisition frequency is set to one data packet per second. The thermal power generating unit dispatching command includes three core parameters: power setpoint, regulation rate, and effective duration. The command parsing engine uses a syntax parse tree structure to handle the differences in command formats specific to different manufacturers. The calculation of the optimal charge and discharge power range of the battery pack is based on the life prediction vector output by the life prediction core engine. The life prediction vector includes the median predicted value of capacity retention rate, the upper and lower boundaries of the prediction range, and the confidence index. The calculation algorithm adopts a constrained optimization model. The objective function seeks to balance the regulation needs of the thermal power unit with the life loss of the battery pack. The constraints include the current available capacity of the battery pack, the maximum charge and discharge rate, and the thermal safety management limit. The optimization model is solved using the interior point method algorithm, and the upper and lower boundaries of the optimal power range that satisfy all constraints are obtained through iterative calculation.

[0084] The dynamic parameter configuration unit monitors system operating status indicators in real time, including CPU utilization, memory usage, and network latency. The multimodal data fusion acquisition module employs fuzzy logic control for its sampling frequency adjustment strategy. The input variables are the system load index and the required prediction accuracy level, while the output variable is the sampling frequency value. The membership function of the fuzzy logic controller uses a trigonometric function form, the rule base contains nineteen control rules, and the defuzzification method uses the centroid method to calculate the accurate output value. The electrochemical state reconstruction module has three accuracy levels: high-precision mode, balanced mode, and fast mode. High-precision mode uses double-precision floating-point arithmetic and fine mesh generation; balanced mode uses single-precision floating-point arithmetic and medium-level mesh generation; and fast mode uses fixed-point arithmetic and coarse mesh generation. The accuracy level switching mechanism is based on the battery state change rate statistically analyzed via a sliding window; when the change rate exceeds a threshold, the accuracy level is automatically increased.

[0085] The instruction verification mechanism of the thermal power generation and energy storage joint commissioning interface module includes dual security measures: digital signature verification and timestamp verification. The digital signature algorithm adopts the ECDSA elliptic curve digital signature algorithm, and the private key is stored in the hardware security module. The dynamic adjustment of the optimal charge / discharge power range of the battery pack considers real-time electricity price signals. Electricity price data is obtained from the electricity market trading platform, and the price factor is converted into a battery pack operating cost item through the price elasticity coefficient. The power range calculation results are published to the battery management system via the OPC-UA protocol, with the data format conforming to the IEEE 1815 standard. The transmission cycle is synchronized with the update cycle of the thermal power generation unit dispatch instructions. The parameter adaptive adjustment process of the dynamic parameter configuration unit adopts a closed-loop control strategy, with a control cycle set to once every five minutes. A system stability assessment is performed before each adjustment. The sampling frequency adjustment range of the multi-modal data fusion acquisition module is continuously adjustable from 1Hz to 1kHz, with an adjustment step size of 1Hz. The frequency change uses a ramp function to avoid abrupt changes. The electrochemical state reconstruction module has a minimum hold time for switching between calculation accuracy levels. The minimum hold time is 10 minutes for high-precision mode, 5 minutes for equilibrium mode, and 2 minutes for fast mode to prevent calculation oscillations caused by frequent switching.

[0086] The communication interruption handling mechanism between the thermal power generation unit and the thermal power storage joint commissioning interface module adopts heartbeat detection technology. Heartbeat packets are sent at three-second intervals, and a communication interruption is determined by the loss of five consecutive heartbeat packets. During a communication interruption, the system switches to local cache mode, using the most recent effective power range value, and simultaneously records the number of lost scheduling instructions and the maximum interruption duration. After communication is restored, a data retransmission program is automatically executed. This program uses differential synchronization to transmit only the data packets missing during the interruption. The system load assessment algorithm of the dynamic parameter configuration unit uses moving average filtering technology, collecting CPU utilization data from the most recent ten cycles to remove the impact of instantaneous fluctuations. The sampling frequency adjustment of the multimodal data fusion acquisition module considers sensor lifespan factors, establishing a correlation model between sampling frequency and sensor lifespan. While meeting monitoring requirements, a lower sampling frequency is selected as much as possible to extend sensor lifespan. The electrochemical state reconstruction module sets a hysteresis range for switching calculation accuracy levels to prevent frequent switching near the switching threshold. The threshold for switching from high-precision mode to balanced mode is 5% lower than the threshold for switching from balanced mode to high-precision mode.

[0087] The command priority management of the thermal power-storage joint commissioning interface module supports conflict resolution from multiple command sources. When different commands from multiple sources are received simultaneously, they are processed according to their priority. The command source priority, from highest to lowest, is: power grid dispatch center, power plant operators, and automatic generation control system. A manual intervention mechanism is introduced for calculating the optimal charge / discharge power range of the battery pack. Operators can manually set power range limits, and the system automatically adds the manually set constraints to the calculation results. The parameter configuration interface of the dynamic parameter configuration unit is developed using web technology, supporting cross-platform access and real-time parameter monitoring. The sampling frequency adjustment records of the multimodal data fusion acquisition module are saved to the operation log, which includes the adjustment time, the frequency value before adjustment, the frequency value after adjustment, and a description of the adjustment reason. The electrochemical state reconstruction module has an early warning mechanism for switching calculation accuracy levels. When switching to fast mode is required, an early warning message is sent to the operators in advance via both SMS and email.

[0088] The performance monitoring indicators of the thermal power-storage joint commissioning interface module include command response time, calculation latency time, and number of communication interruptions. Monitoring data is displayed in real-time on the monitoring screen. The accuracy of the calculated optimal charge / discharge power range for the battery pack is verified using backtesting technology, comparing actual operating data with historical prediction results and calculating the average absolute percentage error as the evaluation indicator. The parameter adjustment effect of the dynamic parameter configuration unit is evaluated every 24 hours, and evaluation reports are automatically generated and archived.

[0089] The data interface between the thermal power generation and energy storage integration interface module and the battery management system adopts a publish-subscribe model. Topic naming follows a hierarchical naming convention: the first level is the power plant number, the second level is the battery pack number, and the third level is the data type. The dynamic parameter configuration unit's anomaly handling mechanism includes parameter out-of-bounds detection. When an abnormal parameter value is detected, it automatically restores to the safe default value and triggers the system diagnostic program. The sampling frequency adjustment of the multimodal data fusion acquisition module considers data quality requirements, automatically increasing the sampling frequency to ensure data validity when data quality deteriorates. The electrochemical state reconstruction module's calculation accuracy level switching uses a gradual transition mechanism, employing a data smoothing algorithm during level switching to avoid abrupt output value changes. The communication protocol converter of the thermal power generation and energy storage integration interface module supports multiple protocol conversions, including common protocol conversions such as IEC-104 to MODBUS and DNP3 to OPC-UA. The calculation of the optimal charge / discharge power range for the battery pack considers the uneven aging of the battery pack, setting stricter power limits for battery cells in poor health. The system parameter backup mechanism of the dynamic parameter configuration unit adopts an incremental backup strategy, performing a full backup once a day at midnight and an incremental backup once an hour, retaining backup data for thirty days. Clock synchronization accuracy must be at the millisecond level, employing the IEEE 1588 precise time protocol to synchronize the clocks of all devices within the system. The dynamic parameter configuration unit uses a Git version control system for parameter version management; each parameter modification generates a new version record, supporting parameter rollback and modification history queries. The multimodal data fusion acquisition module has a maximum rate-of-change limit for sampling frequency adjustment, with a single adjustment not exceeding 20% ​​of the current value to avoid data anomalies caused by sudden changes in sampling frequency. The electrochemical state reconstruction module has a manual locking function for switching calculation accuracy levels, allowing operators to temporarily lock the current accuracy level and prevent automatic switching.

[0090] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0091] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A non-destructive testing and life prediction system for modular lithium battery packs designed for combined thermal and energy storage operation, characterized in that, include: The multimodal data fusion acquisition module uses a distributed fiber optic sensing array and an electromagnetic acoustic emission sensor to simultaneously capture the surface temperature field distribution, internal stress wave propagation characteristics, and partial discharge signals of the lithium battery pack, thereby obtaining multi-dimensional physical information during battery operation and reflecting the coupling state of multiple physical fields inside the battery. The electrochemical state reconstruction module, based on the heterogeneous sensing data output by the multimodal data fusion acquisition module, separates the ohmic polarization, concentration polarization and electrochemical polarization components through the polarization voltage decomposition algorithm, and reconstructs the three-dimensional distribution of ion migration paths inside the battery. The core engine for lifetime prediction uses the polarization components and ion migration paths output by the electrochemical state reconstruction module to establish a mapping relationship between battery capacity decay and multi-physics parameters using a degradation trajectory matching algorithm, generating a lifetime prediction vector that includes cycle number and calendar aging factor. The anomaly propagation and source tracing module maps the anomaly nodes output by the life prediction core engine to the battery pack spatial topology, and combines the thermal-mechanical-electric coupling field simulation to deduce the propagation path of anomalies among multiple battery cells. The health status visualization module integrates lifespan prediction vectors and anomaly propagation path data to generate a hierarchical health status map of the battery pack. The lifespan prediction core engine specifically includes: The degradation feature extraction unit performs principal component analysis on the ion diffusion flux density output by the electrochemical state reconstruction module and selects the top three principal components with the highest correlation to capacity decay as feature vectors. The trajectory matching database stores historical degradation trajectory data of lithium batteries with different material systems under various working conditions. Each trajectory includes capacity decay rate, polarization parameter change trend and temperature stress coefficient. The real-time matching calculation unit compares the feature vector of the current battery pack with the historical data in the trajectory matching database through dynamic time warping, and outputs the five closest reference degradation trajectories and their weight coefficients. The lifetime prediction core engine also includes: The remaining lifetime calculation unit uses a weighted average algorithm to fuse the prediction results of each trajectory based on the reference degradation trajectory output by the real-time matching calculation unit, and generates a lifetime prediction range that includes the median capacity retention rate and upper and lower boundaries. The calendar aging correction unit calculates the impact of temperature acceleration factor on the non-cycle aging portion of the battery based on the Arrhenius equation and dynamically adjusts the boundary values ​​of the lifetime prediction range. The anomaly propagation tracing module specifically includes: The battery pack topology modeling unit constructs a three-dimensional mesh model based on the physical connection method of the modular lithium battery pack, and marks the thermal conductivity coefficient and current path impedance between each battery cell. The multi-field coupled simulation unit uses the abnormal nodes output by the lifetime prediction core engine as the initial excitation source and solves the coupled differential equations of the thermal field and stress field through an implicit algorithm in alternating directions. The propagation path marker unit records the sequence of mesh cells whose abnormal parameters exceed the threshold during the simulation, forming a directed propagation link graph of the abnormality. The anomaly propagation tracing module also includes: The critical state early warning unit generates an early warning signal containing the affected battery number when the abnormal diffusion range output by the propagation path marking unit touches the battery pack boundary unit. The failure mode classification unit matches a library of typical failure modes based on the spatial distribution characteristics of abnormal propagation paths, and outputs a short-circuit risk level or thermal runaway probability index.

2. The modular lithium battery pack non-destructive testing and life prediction system for combined thermal and energy storage as described in claim 1, characterized in that, The multimodal data fusion and acquisition module specifically includes: The distributed fiber optic temperature sensing unit arranges a spiral-wound fiber optic grid along the surface of the battery module and obtains the temperature gradient distribution through Brillouin dispersion radio frequency shift analysis technology. The electromagnetic acoustic emission detection unit uses a wideband piezoelectric transducer array to capture stress wave signals during battery charging and discharging, and uses the wavelet packet energy entropy algorithm to extract the frequency band features of the stress wave. The partial discharge monitoring unit collects internal discharge pulse signals of the battery through an ultra-high frequency sensor and identifies the discharge type by combining the pulse phase resolution spectrum. The data synchronization and alignment unit performs time stamp calibration on temperature field data, stress wave characteristics and discharge pulse signals, and establishes a timestamp alignment mechanism for cross-modal data.

3. The modular lithium battery pack non-destructive testing and life prediction system for combined thermal and energy storage as described in claim 2, characterized in that, The electrochemical state reconstruction module specifically includes: The dynamic impedance calculation unit performs Hilbert transform on the partial discharge signal output by the multi-modal data fusion acquisition module and extracts the discharge pulse envelope as the excitation response input. The polarization component separation unit uses a recursive least squares algorithm to fit the dynamic impedance spectrum curve and decompose it to obtain the ohmic polarization resistance, electrochemical polarization capacitance and concentration polarization time constant. The ion migration modeling unit constructs an equivalent circuit model of the porous electrode based on the parameters output by the polarization component separation unit, and calculates the diffusion flux density of lithium ions in the positive and negative electrode materials using the finite element method.

4. The modular lithium battery pack non-destructive testing and life prediction system for combined thermal and energy storage as described in claim 1, characterized in that, The health status visualization module specifically includes: The 3D rendering engine converts the ion migration path data output by the electrochemical state reconstruction module into a color gradient encoded 3D streamline map. The lifespan degradation heatmap generates color temperature gradient effects for each cell in the battery pack based on the predicted range values ​​output by the lifespan prediction core engine. The anomaly diffusion animation unit generates a dynamic diffusion process simulation image based on the directed propagation link diagram output by the anomaly propagation tracing module.

5. The modular lithium battery pack non-destructive testing and life prediction system for combined thermal and energy storage as described in claim 4, characterized in that, The health status visualization module also includes: The interactive diagnostic unit responds to externally inputted battery cell selection commands and retrieves the historical polarization parameter curves and current ion migration rate data for the corresponding cell. The comparative analysis view displays the differences in health status indicators of similar battery modules at the same number of cycles.

6. The modular lithium battery pack non-destructive testing and life prediction system for combined thermal and energy storage as described in claim 1, characterized in that, Also includes: The thermal power generation and storage joint commissioning interface module receives the dispatching instructions from the thermal power generating unit and calculates the optimal charging and discharging power range of the battery pack based on the life prediction vector. The dynamic parameter configuration unit adjusts the sampling frequency of the multimodal data fusion acquisition module and the calculation accuracy level of the electrochemical state reconstruction module in real time.

Images