Intelligent lithium battery pack multi-cell parallel management system and method

Abstract

The application belongs to the technical field of battery pack parallel control, and discloses an intelligent lithium battery pack multi-battery parallel management system and method; by collecting the transient charge distribution and electrode interface impedance data of each battery unit in real time, analyzing the charge migration rate and impedance distribution characteristics, accurately extracting the dynamic offset characteristics of the parasitic capacitance, and identifying the core influence path of the parasitic capacitance on the current distribution, a parasitic capacitance nonlinear coupling model is constructed for predicting and optimizing the current transient distribution. At the same time, by monitoring the time domain distortion characteristics of the charge pulse signal, the current distribution compensation strategy is dynamically adjusted, so as to realize the accurate balance of the current among the battery units. The application improves the performance, safety and service life of the battery pack in the high-frequency charging and discharging scene, reduces the maintenance cost, and reduces environmental pollution.

Description

Technical Field

[0001] This invention relates to the field of parallel control technology for battery packs, and more specifically, to a smart lithium battery pack multi-cell parallel management system and method. Background Technology

[0002] With the global acceleration of energy transition and electrification, lithium-ion batteries, with their high energy density, long cycle life, and environmentally friendly characteristics, have become the core power source for electric transportation, renewable energy storage, and portable electronic devices. In large-scale battery systems, multi-cell parallel architectures are widely used due to their advantages such as strong scalability and high fault tolerance. Traditional parallel battery management systems mainly employ impedance matching-based balancing techniques, such as active balancing and passive balancing methods.

[0003] However, in existing parallel lithium battery pack management systems, the problems caused by inconsistent parasitic capacitance are particularly prominent in high-frequency charge and discharge scenarios, becoming a key bottleneck restricting battery pack performance, safety, and lifespan. Specifically, due to differences in manufacturing processes, material characteristics, and aging levels among battery cells in a parallel battery pack, the parasitic capacitance of each cell often exhibits significant inconsistency. This inconsistency can cause transient uneven current distribution during high-frequency charge and discharge. For example, in scenarios such as fast charging or high-power discharging (e.g., rapid acceleration) of electric vehicles, battery cells with larger parasitic capacitance tend to absorb or release less current due to their slower charge migration rate, while cells with smaller parasitic capacitance are subjected to excessively high current surges, leading to overcharging or over-discharging of some battery cells and thus the risk of thermal runaway. Furthermore, existing technologies are insufficient in monitoring the dynamic changes of parasitic capacitance, making it difficult to capture the periodic fluctuations and shift characteristics of parasitic capacitance under high-frequency excitation in real time. Traditional linear models also cannot accurately describe the nonlinear coupling relationship between parasitic capacitance and impedance distribution and charge migration. For example, in practical applications such as high-frequency pulse discharge in drones or rapid load switching in energy storage systems, the transient current unevenness caused by inconsistent parasitic capacitance is often simplified to a static impedance difference problem, ignoring the complex influence path of electrode interface impedance abrupt changes and their local topological characteristics on current distribution. This lack of analysis makes it impossible for existing systems to accurately predict and dynamically adjust current distribution, leading to a decrease in overall battery pack efficiency, accelerated aging, and even safety accidents caused by local overheating in extreme cases.

[0004] In view of this, the present invention proposes a smart lithium battery pack multi-cell parallel management system and method to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a smart lithium battery pack multi-cell parallel management system, comprising:

[0006] The data acquisition module is used to collect transient charge distribution data and electrode interface impedance data of each battery cell in the parallel lithium battery pack during high-frequency charging and discharging.

[0007] The charge migration analysis module is used to calculate the charge migration rate of each battery cell during high-frequency charging and discharging based on transient charge distribution data, and to extract the dynamic offset characteristics of parasitic capacitance of each battery cell based on the charge migration rate.

[0008] The impedance network modeling module is used to construct the impedance distribution network model of each battery cell based on the electrode interface impedance data, and to identify the core influence path of parasitic capacitance inconsistency through local topological analysis of impedance abrupt change points in the impedance distribution network model.

[0009] The coupling model construction module is used to construct a nonlinear coupling model of parasitic capacitance based on the dynamic offset characteristics and core influence paths of parasitic capacitance during high-frequency charging and discharging.

[0010] The distortion feature extraction module is used to extract the distortion components related to the parasitic capacitance nonlinear coupling model in the charge pulse signal by real-time monitoring of the time-domain distortion characteristics of the charge pulse signal of each battery cell.

[0011] The dynamic adjustment module is used to dynamically adjust the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model based on the time-domain offset and amplitude attenuation rate of the distortion component.

[0012] The charge / discharge control module is used to control the charge / discharge management equipment of the parallel lithium battery pack based on the parameters of the adjusted parasitic capacitance nonlinear coupling model, so as to achieve precise balance of transient current distribution among the battery cells.

[0013] Preferably, based on transient charge distribution data, the charge migration rate of each battery cell during high-frequency charging and discharging is calculated, and the dynamic shift characteristics of the parasitic capacitance of each battery cell are extracted based on the charge migration rate, including:

[0014] The transient charge distribution data is segmented in the time domain to obtain the charge accumulation in each time domain segment, which is denoted as the segmented charge amount.

[0015] Calculate the difference in charge between adjacent time domain segments, and combine this with the duration of the time domain segments to calculate the charge migration rate of each battery cell.

[0016] A sliding window analysis was performed on the charge migration rate to extract the periodic fluctuation component of the charge migration rate under high-frequency charge and discharge excitation, which was denoted as the charge migration fluctuation characteristic.

[0017] Based on the amplitude and phase changes of charge migration fluctuation characteristics, and combined with a preset parasitic capacitance reference model, the dynamic offset characteristics of parasitic capacitance of each battery cell are calculated. The dynamic offset characteristics of parasitic capacitance are determined by the deviation between the amplitude of charge migration fluctuation characteristics and the parasitic capacitance reference model.

[0018] Preferably, based on the electrode interface impedance data, an impedance distribution network model for each battery cell is constructed, and the core influencing path of parasitic capacitance inconsistency is identified through local topological analysis of impedance abrupt change points in the impedance distribution network model, including:

[0019] The electrode interface impedance data is normalized to obtain the normalized impedance value of each battery cell.

[0020] An impedance distribution network model is constructed using each battery cell as a node and the difference in normalized impedance between adjacent battery cells as the edge weight.

[0021] In the impedance distribution network model, the local standard deviation of the edge weights of each node is calculated and denoted as the local impedance variability.

[0022] Nodes whose local impedance fluctuation exceeds a preset fluctuation threshold are marked as impedance abrupt change points, and the set of edges directly connected to these impedance abrupt change points is extracted and denoted as a local topological subgraph.

[0023] Shortest path analysis is performed on the local topology subgraph to identify the shortest path connecting impedance abrupt changes and mark the shortest path as the core influencing path.

[0024] Preferably, based on the dynamic offset characteristics and core influence paths of parasitic capacitance, a nonlinear coupling model of parasitic capacitance for transient current distribution during high-frequency charging and discharging is constructed, including:

[0025] Based on the dynamic offset characteristics of parasitic capacitance, the parasitic capacitance offset of each battery cell is calculated, and the node attributes of the impedance distribution network model are updated with the parasitic capacitance offset as the weight to generate a parasitic capacitance weighted network model.

[0026] In the parasitic capacitance weighted network model, the set of edges corresponding to the core influence path is labeled to generate a parasitic capacitance influence subgraph containing the core influence path;

[0027] Based on the parasitic capacitance influence subgraph, the current transient distribution interference factor of each edge on the core influence path is calculated. The current transient distribution interference factor is determined by the nonlinear combination of the parasitic capacitance offset of the edge and the impedance value of the edge.

[0028] Based on the global topological characteristics of the current transient distribution interference factor and the parasitic capacitance weighted network model, a parasitic capacitance nonlinear coupling model is constructed. The parasitic capacitance nonlinear coupling model includes the predicted value of the current transient distribution and the compensation requirement value of each battery cell.

[0029] Preferably, by real-time monitoring of the time-domain distortion characteristics of the charge pulse signals of each battery cell, the distortion components related to the nonlinear coupling model of parasitic capacitance in the charge pulse signals are extracted, including:

[0030] The charge pulse signals of each battery cell are collected, and the charge pulse signals are sampled in the time domain to obtain discrete time series data;

[0031] Calculate the standard pulse template of the charge pulse signal, and compare the actual collected charge pulse signal with the standard pulse template to obtain the difference signal;

[0032] The difference signal is subjected to threshold judgment. When the amplitude of the difference signal exceeds the preset distortion threshold, it is marked as a distortion moment.

[0033] The time interval between adjacent distortion time points is statistically analyzed. When the deviation between the time interval and the characteristic period of the parasitic capacitance nonlinear coupling model is less than a preset deviation threshold, the corresponding difference signal segment is extracted as a distortion component.

[0034] Preferably, the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model is dynamically adjusted based on the time-domain offset and amplitude attenuation rate of the distortion component, including:

[0035] Determine the start time of the distorted component and calculate its time difference with the start time of the ideal pulse signal to obtain the time domain offset;

[0036] Measure the peak amplitude and the end amplitude of the distortion component, and calculate the amplitude attenuation rate, which is equal to the difference between the peak amplitude and the end amplitude divided by the distortion duration.

[0037] A first compensation coefficient is set based on the time domain offset. When the time domain offset is greater than a first threshold, the first compensation coefficient is increased; when the time domain offset is less than a second threshold, the first compensation coefficient is decreased.

[0038] A second compensation coefficient is set based on the amplitude attenuation rate. When the amplitude attenuation rate is greater than the attenuation threshold, the adjustment range of the second compensation coefficient is increased.

[0039] The first compensation coefficient and the second compensation coefficient are weighted and combined to obtain the final current distribution compensation coefficient, and the corresponding parameters of the parasitic capacitance nonlinear coupling model are updated.

[0040] Preferably, a sliding window analysis is performed on the charge migration rate to extract the periodic fluctuation component of the charge migration rate under high-frequency charge-discharge excitation, which is denoted as the charge migration fluctuation characteristic, including:

[0041] The charge migration rate is divided into sliding windows with a preset window length to obtain the charge migration rate subsequence within each sliding window;

[0042] Autocorrelation analysis was performed on the charge mobility rate subsequence to calculate the autocorrelation coefficient sequence of the charge mobility rate subsequence;

[0043] Peak points are extracted from the autocorrelation coefficient sequence, and the time interval between adjacent peak points is calculated and denoted as the fluctuation period.

[0044] The periodic fluctuation component of the charge migration rate is determined by the ratio of the fluctuation period to the high-frequency charge-discharge excitation frequency, and is denoted as the charge migration fluctuation characteristic.

[0045] Preferably, shortest path analysis is performed on the local topology subgraph to identify the shortest path connecting impedance abrupt changes, and the shortest path is marked as the core influencing path, including:

[0046] In the local topological subgraph, starting from the impedance abrupt change point, the path length to each adjacent node is calculated. The path length is determined by weighting the normalized impedance values ​​of the edges.

[0047] Based on path length, Dijkstra's algorithm is used to calculate the shortest path from the impedance abrupt change point to all other nodes in the local topological subgraph;

[0048] Extract the path with the smallest sum of edge weights from the shortest paths, and denote it as the candidate core path;

[0049] Impedance fluctuation verification is performed on candidate core paths. The mean value of the local impedance fluctuation of each side on the candidate core path is calculated. Candidate core paths with a mean value greater than a preset verification threshold are marked as core influence paths.

[0050] Preferably, based on the global topological characteristics of the current transient distribution interference factor and the parasitic capacitance weighted network model, a parasitic capacitance nonlinear coupling model is constructed, including:

[0051] Calculate the global clustering coefficients of the parasitic capacitance weighted network model, and denote them as global topological feature values;

[0052] The interference propagation weights of each node in the parasitic capacitance weighted network model are determined by multiplying the global topological eigenvalues ​​with the current transient allocation interference factor.

[0053] Based on the interference propagation weight, a weighted adjacency matrix is ​​constructed for the parasitic capacitance weighted network model;

[0054] A nonlinear transformation is performed on the weighted adjacency matrix to generate a parasitic capacitance nonlinear coupling model. The parasitic capacitance nonlinear coupling model determines the predicted value of the transient current distribution and the compensation requirement value of each battery cell through the eigenvalue decomposition of the weighted adjacency matrix.

[0055] A method for parallel management of multiple cells in an intelligent lithium battery pack, implemented based on an intelligent lithium battery pack parallel management system, includes:

[0056] Step 1: Collect transient charge distribution data and electrode interface impedance data of each battery cell in the parallel lithium battery pack during high-frequency charging and discharging.

[0057] Step 2: Based on the transient charge distribution data, calculate the charge migration rate of each battery cell during the high-frequency charge and discharge process, and extract the dynamic offset characteristics of the parasitic capacitance of each battery cell based on the charge migration rate.

[0058] Step 3: Based on the electrode interface impedance data, construct the impedance distribution network model of each battery cell, and identify the core influence path of parasitic capacitance inconsistency through local topological analysis of impedance abrupt change points in the impedance distribution network model.

[0059] Step 4: Based on the dynamic offset characteristics and core influence path of parasitic capacitance, construct a nonlinear coupling model of parasitic capacitance for transient current distribution during high-frequency charging and discharging.

[0060] Step 5: By real-time monitoring of the time-domain distortion characteristics of the charge pulse signals of each battery cell, extract the distortion components in the charge pulse signals that are related to the nonlinear coupling model of parasitic capacitance.

[0061] Step 6: Dynamically adjust the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model based on the time-domain offset and amplitude attenuation rate of the distortion component.

[0062] Step 7: Based on the parameters of the adjusted parasitic capacitance nonlinear coupling model, control the charging and discharging management equipment of the parallel lithium battery pack.

[0063] The technical effects and advantages of the intelligent lithium battery pack multi-cell parallel management system and method of the present invention are as follows:

[0064] This invention improves the performance, safety, and lifespan of parallel lithium battery packs in high-frequency charge and discharge scenarios. In fast charging or high-power discharging scenarios, the battery pack achieves more stable operation, avoiding the risk of thermal runaway caused by overcharging or over-discharging of individual battery cells, thus improving safety and reliability. For devices requiring high-frequency pulse discharge, the overall efficiency of the battery pack is optimized, extending the driving range and enhancing the stability of device operation. In energy storage systems, especially in scenarios with rapid load switching, the battery pack can more efficiently handle transient current surges, reducing energy loss and extending device lifespan. Furthermore, this invention reduces maintenance and replacement costs by effectively mitigating aging differences between battery cells. Attached Figure Description

[0065] Figure 1 This is a schematic diagram of a multi-cell parallel management method for an intelligent lithium battery pack according to the present invention;

[0066] Figure 2 This is a schematic diagram of a smart lithium battery pack multi-cell parallel management system according to the present invention. Detailed Implementation

[0067] 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.

[0068] This application provides an intelligent lithium battery pack multi-cell parallel management system and method. The execution entities of the system include, but are not limited to, battery management systems (BMS), energy storage controllers, electric vehicle battery monitoring devices, and intelligent charging station management systems, which can be regarded as general computing nodes of this application. The intelligent management system includes, but is not limited to, at least one of cloud-based battery monitoring platforms, distributed battery balancing systems, and intelligent charge and discharge controllers.

[0069] Please see Figure 1 In this embodiment of the invention, the specific implementation process of a method for managing multiple batteries in a smart lithium battery pack in parallel includes:

[0070] Step 1: Collect transient charge distribution data and electrode interface impedance data of each battery cell in the parallel lithium battery pack during high-frequency charging and discharging. Transient charge distribution data and electrode interface impedance data are core monitoring indicators of the battery management system, collected in real time through a high-precision sensor network. Transient charge distribution data directly reflects the dynamic characteristics of charge migration within the battery, while electrode interface impedance data characterizes the impedance changes during the battery's electrochemical reaction process. These data provide a foundation for analyzing parasitic capacitance characteristics and formulating current distribution strategies, ensuring the targetedness and effectiveness of the management system.

[0071] Step 2: Based on the transient charge distribution data, calculate the charge migration rate of each battery cell during high-frequency charge and discharge, and extract the dynamic shift characteristics of parasitic capacitance for each battery cell based on the charge migration rate. The charge migration rate quantifies the dynamic characteristics of charge transport within the battery and directly determines the behavior of parasitic capacitance. The dynamic shift characteristics of parasitic capacitance reflect the changes in capacitance parameters of the battery cell under high-frequency operating conditions, providing key parameters for subsequent model construction. These two indicators are obtained through time-domain segmentation and fluctuation characteristic analysis of the transient charge distribution data, ensuring that the current distribution strategy matches the actual electrochemical characteristics of the battery.

[0072] Step 3: Based on the electrode interface impedance data, construct an impedance distribution network model for each battery cell, and identify the core influencing paths of parasitic capacitance inconsistencies through local topological analysis of impedance abrupt change points in the impedance distribution network model. This module first normalizes the electrode interface impedance data and constructs an impedance distribution network model with battery cells as nodes. Then, through accurate local fluctuation calculation and topological analysis, it identifies impedance abrupt change points in the network and their associated core influencing paths, laying the foundation for subsequent coupled model construction.

[0073] Step 4: Based on the dynamic offset characteristics and core influence paths of parasitic capacitance, a nonlinear coupling model of parasitic capacitance for transient current distribution during high-frequency charging and discharging is constructed. This module calculates the transient current distribution interference factor on the core influence path by constructing a parasitic capacitance weighted network model, establishing a nonlinear mapping relationship between current distribution and parasitic capacitance characteristics, and achieving accurate modeling of the internal current distribution mechanism of the battery pack, providing a theoretical basis for subsequent equalization control.

[0074] Step 5: By real-time monitoring of the time-domain distortion characteristics of the charge pulse signals of each battery cell, the distortion components related to the parasitic capacitance nonlinear coupling model in the charge pulse signals are extracted. This module identifies the distortion components that match the frequency of the parasitic capacitance coupling characteristics through Hilbert transform analysis of the charge pulse signals, thereby achieving accurate monitoring of the battery operating status and providing real-time basis for dynamic adjustment of model parameters.

[0075] Step 6: Dynamically adjust the current distribution compensation coefficients in the parasitic capacitance nonlinear coupling model based on the time-domain offset and amplitude attenuation rate of the distortion component. This module establishes a mapping relationship between distortion characteristics and compensation coefficients by accurately calculating the time-domain characteristic parameters of the distortion component, thereby achieving adaptive adjustment of model parameters and ensuring the real-time effectiveness of the current distribution strategy.

[0076] Step 7: Based on the parameters of the adjusted parasitic capacitance nonlinear coupling model, control the charge / discharge management device of the parallel lithium battery pack to achieve precise balance in the transient current distribution among the battery cells. This module converts the optimized model parameters into specific control commands, driving the charge / discharge management device to execute a precise current distribution strategy, ensuring a balanced charge / discharge state among the cells in the battery pack, and improving the overall performance and lifespan of the battery pack.

[0077] In this embodiment of the invention, the detailed implementation steps for calculating the charge migration rate of each battery cell during high-frequency charging and discharging based on transient charge distribution data, and extracting the dynamic offset characteristics of the parasitic capacitance of each battery cell based on the charge migration rate, include:

[0078] Transient charge distribution data is processed in the time domain to obtain the accumulated charge within each time domain segment, denoted as the segmented charge. Time-domain segmentation eliminates the complexity of continuous signals, discretizing the charge distribution into calculable segmented values ​​for easier subsequent analysis. The processing employs an equal-time-interval sampling method, determining sampling points based on the charge-discharge cycle; each sampling point corresponds to the charge state of a time domain segment. High-frequency noise during sampling is preprocessed using wavelet denoising techniques to ensure the reliability and representativeness of the sampled data. The segmented charge is calculated through time-domain integration, accurately reflecting the accumulated charge state within each time window.

[0079] The difference in charge between adjacent time-domain segments is calculated, and combined with the duration of each time-domain segment, the charge migration rate of each battery cell is calculated. Charge migration rate is a core indicator characterizing the dynamic behavior of charge within the battery, directly reflecting the battery's charge-discharge response characteristics. The calculation formula is:

[0080] ;

[0081] in, For the first Each battery cell The charge transfer rate at time t, for The piecewise charge at time t, for The piecewise charge at time t, The duration of the time-domain segment.

[0082] The charge migration rate is measured in coulombs per second, which is equivalent to current but more accurately reflects the dynamics of microscopic charge migration. This metric provides fundamental data for subsequent analysis of parasitic capacitance characteristics and offers a time-domain characterization of battery response properties.

[0083] A sliding window analysis was performed on the charge migration rate to extract the periodic fluctuation components of the charge migration rate under high-frequency charge-discharge excitation, denoted as the charge migration fluctuation characteristics. Sliding window analysis is an effective method for discovering periodic patterns in charge migration, capable of extracting implicit periodic features from time-series data. The analysis process uses an integer multiple of the high-frequency charge-discharge cycle as the window length to slide and segment the charge migration rate sequence, identifying the periodic components through autocorrelation analysis. The periodic fluctuation components directly reflect the dynamic response characteristics of the battery under high-frequency operating conditions, providing a basis for extracting parasitic capacitance parameters.

[0084] Based on the amplitude and phase changes of charge migration fluctuation characteristics, and combined with a pre-defined parasitic capacitance benchmark model, the dynamic offset characteristics of parasitic capacitance for each battery cell are calculated. These dynamic offset characteristics are determined by the deviation between the amplitude of the charge migration fluctuation characteristics and the parasitic capacitance benchmark model. The dynamic offset characteristics of parasitic capacitance are a key indicator describing the changes in battery capacitance parameters under high-frequency operating conditions, directly affecting the current distribution balance. The calculation employs a frequency domain mapping method to establish the mapping relationship between charge migration fluctuation characteristics and parasitic capacitance parameters. The parasitic capacitance benchmark model is constructed based on the theoretical characteristics of the battery and historical operating data, providing standard reference values. The offset characteristics are obtained by comparing the actual fluctuation characteristics with the benchmark model, accurately reflecting the dynamic change law of the battery's parasitic capacitance and providing key parameters for subsequent coupled model construction.

[0085] In this embodiment of the invention, the detailed implementation steps for constructing an impedance distribution network model for each battery cell based on electrode interface impedance data, and identifying the core influencing path of parasitic capacitance inconsistency through local topological analysis of impedance abrupt change points in the impedance distribution network model, include:

[0086] The electrode interface impedance data were normalized to obtain the normalized impedance value of each battery cell. Normalization is a key preprocessing step to eliminate dimensional differences and improve data comparability. The processing employed a maximum-minimum normalization method. Normalization not only unified the data scale but also facilitated subsequent network construction and analysis, ensuring the comparability of impedance characteristics between different battery cells.

[0087] An impedance distribution network model is constructed, using each battery cell as a node and the difference in normalized impedance between adjacent battery cells as the edge weight. Network model construction is the core step in revealing the impedance relationships between battery cells. Graph theory methods are used to transform discrete impedance data into a relational network with a topological structure. In the construction process, each battery cell is considered a network node, and the connections between nodes are determined based on the physical layout and electrical connections. The edge weights are calculated using the normalized impedance difference.

[0088] ;

[0089] in, For connecting nodes and The edge weights represent the degree of impedance difference between the two battery cells. For the first The normalized impedance value of each battery cell. For the first The normalized impedance value of each battery cell.

[0090] The network model visually illustrates the impedance distribution within the battery pack, laying the foundation for subsequent unbalanced path identification. Edges with larger weights indicate more significant impedance differences between connected battery cells, and are more likely to become critical paths of imbalance.

[0091] In an impedance distribution network model, the local standard deviation of the edge weights for each node is calculated and denoted as impedance local volatility. Impedance local volatility is a key indicator for identifying impedance abrupt changes, reflecting the degree of non-uniformity in impedance distribution around a node. The calculation process first determines the set of adjacent edges for each node, then calculates the standard deviation of these edge weights to obtain the local volatility value. Higher volatility indicates a more non-uniform impedance distribution around the node, making it more likely to be a critical point affecting current distribution balance. This indicator transforms the global structural characteristics of the network into a description of the local characteristics of the nodes, facilitating the precise location of key nodes.

[0092] Nodes with local impedance fluctuations exceeding a preset fluctuation threshold are marked as impedance abrupt change points. The set of edges directly connected to these abrupt change points is extracted, denoted as a local topological subgraph. Impedance abrupt change points are critical locations of impedance anomalies, directly affecting the balance of current distribution. The marking process employs an adaptive threshold method, determining a reasonable preset fluctuation threshold based on the global fluctuation distribution, typically the upper quartile of the global fluctuation. After marking the abrupt change points, all directly connected edges are extracted using an adjacency matrix, forming a local topological subgraph. This subgraph contains core path information that may affect the balance of current distribution, providing input for subsequent shortest path analysis.

[0093] Shortest path analysis is performed on the local topological subgraph to identify the shortest paths connecting points of impedance abrupt changes, and these shortest paths are marked as core influencing paths. Shortest path analysis is the core algorithm for determining key influencing paths, using graph theory to find the optimal connection paths between abrupt change points. Dijkstra's algorithm is employed, using edge weights as path costs to calculate the shortest connection paths between abrupt change points. The resulting core influencing paths are the most significant propagation channels of parasitic capacitance inconsistency, directly determining the main direction of current distribution imbalance and providing key topological constraints for subsequent coupling model construction.

[0094] In this embodiment of the invention, the detailed implementation steps for constructing a parasitic capacitance nonlinear coupling model for transient current distribution during high-frequency charging and discharging, based on the dynamic offset characteristics and core influence paths of parasitic capacitance, include:

[0095] Based on the dynamic offset characteristics of parasitic capacitance, the parasitic capacitance offset of each battery cell is calculated, and the node attributes of the impedance distribution network model are updated with the parasitic capacitance offset as the weight to generate a parasitic capacitance weighted network model.

[0096] Specifically, the dynamic offset characteristics of parasitic capacitance are normalized to eliminate the influence of dimensions; the deviation value relative to the benchmark model is calculated. :

[0097] ;

[0098] in, For the first The baseline model prediction for each battery cell, For the first Normalized parasitic capacitance offset characteristics of individual battery cells;

[0099] Apply nonlinear transformation to amplify the deviation value Based on the amplified deviation value Calculate the final parasitic capacitance offset :

[0100] ;

[0101] in, This is the scaling factor (usually ranging from 0.1 to 1.0). The aging effect coefficient (usually ranging from 0.5 to 1.5); It is the battery aging factor, which can be derived from the number of charge-discharge cycles and the capacity decay rate.

[0102] The process of updating the node properties of the impedance distribution network model is as follows:

[0103] Comprehensive weight of computing nodes It integrates impedance characteristics and parasitic capacitance characteristics:

[0104] ;in, For the first The normalized impedance value of each battery cell. For nodes The local standard deviation (reflecting the degree of unevenness of the surrounding environment); and Weighting coefficients (satisfying) ,generally , ); This is the local topology adjustment coefficient (usually ranging from 0.2 to 0.8).

[0105] Updating the adjacency matrix of the network model based on comprehensive weights:

[0106] ;

[0107] in, This is the original adjacency matrix. This is the updated adjacency matrix; and They are nodes and The overall weight;

[0108] Then calculate the topological importance of the nodes. :

[0109] Node topological importance reflects the change in a node's relative influence in the network after an update, i.e., node attributes.

[0110] The parasitic capacitance weighted network model is a comprehensive model that integrates impedance and capacitance characteristics, accurately describing the electrical characteristic distribution within the battery pack. The calculation process first converts the dynamic offset characteristics of parasitic capacitance into standardized offset values, which are then added as new attribute values ​​to the original impedance distribution network. The updated network model retains the original impedance topology while adding parasitic capacitance information, comprehensively describing the electrical characteristic distribution of the battery pack and providing a foundation for subsequent interference factor calculations.

[0111] In the parasitic capacitance weighted network model, the set of edges corresponding to the core influencing paths is labeled, generating a parasitic capacitance influence subgraph containing these paths. This subgraph is a refined model of the critical influencing paths, focusing on those most likely to cause uneven current distribution. The generation process uses matrix operations to extract the edges corresponding to the core influencing paths from the parasitic capacitance weighted network, forming a subgraph structure focused on the critical paths. This subgraph contains impedance abrupt change points, connection paths, and their associated parasitic capacitance characteristics, providing a focused analytical object for accurately calculating current distribution interference factors.

[0112] Based on the parasitic capacitance influence subgraph, the transient current distribution interference factor of each edge on the core influence path is calculated. The transient current distribution interference factor is determined by a nonlinear combination of the parasitic capacitance offset and the impedance value of the edge. The transient current distribution interference factor is a key indicator for quantifying the impact of parasitic capacitance on current distribution, directly reflecting the intensity of the imbalance effect. The calculation employs a nonlinear combination method, comprehensively considering the interaction between parasitic capacitance offset and impedance characteristics. The calculation formula is as follows:

[0113] ;

[0114] in, For connecting nodes The transient current distribution interference factor of the edge. This is the normalized impedance difference. and For nodes and The parasitic capacitance offset, and These are nonlinear adjustment parameters (typically taken as 2-3 and 0.5-1.5 respectively).

[0115] This nonlinear combination formula reflects the nonlinear amplification effect of parasitic capacitance offset on current distribution, especially when the impedance difference is small, the impact of parasitic capacitance inconsistency is more significant. The calculation results of the interference factor provide key quantitative parameters for the subsequent construction of the coupling model.

[0116] Based on the global topological characteristics of the current transient distribution interference factor and the parasitic capacitance weighted network model, a parasitic capacitance nonlinear coupling model is constructed. This model includes the predicted current transient distribution and compensation requirements for each battery cell. The parasitic capacitance nonlinear coupling model is the core decision-making basis of the battery management system, accurately describing the influence mechanism of parasitic capacitance characteristics on current distribution. The construction process first calculates the global topological characteristic values ​​of the network and quantifies the overall connectivity characteristics of the network through clustering coefficients. Then, it combines the current interference factor to generate a weighted adjacency matrix. Finally, through matrix eigenvalue decomposition, the predicted current distribution and required compensation values ​​for each battery cell are obtained. This model not only considers local electrical characteristic differences but also integrates global topological information, achieving accurate modeling of the current distribution mechanism and providing a theoretical basis for subsequent control strategies.

[0117] In this embodiment of the invention, the detailed implementation steps for extracting the distortion components related to the parasitic capacitance nonlinear coupling model in the charge pulse signal by real-time monitoring of the time-domain distortion characteristics of the charge pulse signal of each battery cell include:

[0118] Charge pulse signals from each battery cell are acquired and time-domain sampled to obtain discrete-time series data. A high-speed data acquisition card is used, with the sampling frequency set to 10-20 times the charge / discharge frequency to ensure accurate capture of transient signal characteristics. The discrete-time series data contains complete time-domain information of the charge pulse signals, providing a raw data foundation for subsequent distortion feature analysis. The acquisition system is equipped with multi-channel synchronous sampling to ensure time consistency of signal acquisition from each battery cell and eliminate phase deviations between channels. After data acquisition, the raw signals are preprocessed, including removing DC bias and filtering high-frequency noise, to improve signal quality.

[0119] A standard pulse template for the charge pulse signal is calculated, and the actual collected charge pulse signal is compared with the standard pulse template to obtain the difference signal. The standard pulse template is a reference benchmark for the charge pulse signal under ideal conditions, obtained by statistically averaging pulse signals under historical normal operating conditions. The calculation process first collects data from multiple charge and discharge cycles during normal battery operation, and averages the signal values ​​at in-phase points to form the standard pulse template. The comparison process uses point-by-point difference calculation, subtracting the amplitude of the actual signal from the template signal at corresponding moments to obtain the difference signal sequence. The difference signal directly reflects the degree of deviation of the actual signal from the ideal state and is the basic data for identifying distortion characteristics.

[0120] A threshold judgment is performed on the difference signal. When the amplitude of the difference signal exceeds a preset distortion threshold, it is marked as a distortion moment. The preset distortion threshold is set considering the signal noise level and system tolerance requirements, and is typically set to 5%-10% of the standard pulse template amplitude. The threshold judgment uses a sliding window method, calculating the local statistical characteristics of the difference signal within each time window and dynamically adjusting the judgment threshold to improve the accuracy of distortion detection. When the amplitude of the difference signal exceeds the threshold, this moment is recorded as a distortion moment, and the polarity of the distortion (positive or negative) is marked. The distribution characteristics of the distortion moment points reflect the abnormal change patterns of the internal electrical characteristics of the battery pack.

[0121] The time intervals between adjacent distortion points are statistically analyzed. When the deviation of the time interval from the characteristic period of the parasitic capacitance nonlinear coupling model is less than a preset deviation threshold, the corresponding difference signal segment is extracted as the distortion component. The characteristic period is a key time parameter extracted from the parasitic capacitance nonlinear coupling model, reflecting the periodic influence of parasitic capacitance on the signal. The statistical process calculates the time intervals of all adjacent distortion points, forming an interval sequence. The preset deviation threshold is typically set to 10%-20% of the characteristic period to ensure that the extracted distortion component is highly correlated with the parasitic capacitance characteristics. When the time interval meets the deviation requirement, the complete difference signal within that time interval is extracted as the distortion component. This time-correlation-based extraction method effectively filters out random noise interference, ensuring that the extracted distortion component has a clear physical meaning.

[0122] In this embodiment of the invention, the detailed implementation steps for dynamically adjusting the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model based on the time-domain offset and amplitude attenuation rate of the distortion component include:

[0123] The start time of the distortion component is determined, and its time difference with the start time of the ideal pulse signal is calculated to obtain the time-domain offset. The start time is determined using a threshold detection method; when the amplitude of the distortion component first exceeds a noise threshold (usually set to 5% of the maximum amplitude), it is marked as the start time. The start time of the ideal pulse signal is the trigger time of the charge / discharge control signal and serves as a time reference. The formula for calculating the time-domain offset is: Time-domain offset = Start time of distortion component - Start time of ideal pulse signal. A positive value indicates that the distortion lags behind the ideal signal, while a negative value indicates that the distortion leads the ideal signal. The magnitude of the time-domain offset directly reflects the response delay characteristics of the battery cell and is an important basis for adjusting the current distribution timing.

[0124] The peak amplitude and the amplitude at the end of the distortion component are measured, and the amplitude decay rate is calculated. The amplitude decay rate is equal to the difference between the peak amplitude and the amplitude at the end of the distortion component divided by the distortion duration. The peak amplitude is determined by obtaining the maximum value through traversing the distortion component sequence. The amplitude at the end of the distortion component is defined as the amplitude value when the distortion component drops to the noise level (below 5% of the maximum amplitude). The distortion duration is the length of time from the start time to the end time. The formula for calculating the amplitude decay rate is:

[0125] Amplitude decay rate = (peak amplitude - amplitude at the end) / distortion duration. This parameter, measured in amplitude per time, reflects the rate at which the distortion energy dissipates. A larger decay rate indicates a stronger ability of the system to suppress disturbances, while a smaller decay rate indicates a longer duration of distortion effects, requiring more robust compensation measures.

[0126] A first compensation coefficient is set based on the time domain offset. When the time domain offset is greater than a first threshold, the first compensation coefficient is increased; when the time domain offset is less than a second threshold, the first compensation coefficient is decreased. The setting of the first and second thresholds is based on the system's time response requirements. Typically, the first threshold is set to 120% of the ideal response time, and the second threshold is set to 80% of the ideal response time.

[0127] The adjustment of the first compensation coefficient uses a piecewise linear function:

[0128] When the time domain offset is greater than the first threshold, the first compensation coefficient = the base value × (1 + 0.1 × (time domain offset - first threshold) / first threshold);

[0129] When the time domain offset is between the second threshold and the first threshold, the first compensation coefficient equals the baseline value.

[0130] When the time domain offset is less than the second threshold, the first compensation coefficient = the baseline value × (1 - 0.1 × (second threshold - time domain offset) / second threshold).

[0131] This segmented adjustment strategy ensures that the compensation is targeted and effective.

[0132] A second compensation coefficient is set based on the amplitude attenuation rate. When the amplitude attenuation rate exceeds the attenuation threshold, the adjustment range of the second compensation coefficient is increased. The attenuation threshold is set according to the dynamic characteristics of the battery pack, typically 150% of the average attenuation rate under normal operating conditions. The setting logic of the second compensation coefficient is as follows: when the amplitude attenuation rate is greater than the attenuation threshold, it indicates good system stability, and the second compensation coefficient = reference value × 0.8; when the amplitude attenuation rate is less than the attenuation threshold, it indicates a long distortion duration, and the second compensation coefficient = reference value × (1 + 0.2 × (attenuation threshold - amplitude attenuation rate) / attenuation threshold). This setting method allows the system to strengthen the compensation when the distortion duration is long and appropriately reduce the compensation when the system is stable, avoiding over-adjustment.

[0133] The first and second compensation coefficients are weighted and combined to obtain the final current distribution compensation coefficient, which is then used to update the corresponding parameters of the parasitic capacitance nonlinear coupling model. The weighting combination uses a linear weighting method, calculated as follows: Current distribution compensation coefficient = a1 × First compensation coefficient + a2 × Second compensation coefficient, where a1 and a2 are weighting coefficients, satisfying a1 + a2 = 1. The weighting coefficients are set considering the relative importance of time-domain and amplitude characteristics; typically, a1 and a2 are both between 0.4 and 0.6. The calculated current distribution compensation coefficient is directly updated in the parasitic capacitance nonlinear coupling model, replacing the original compensation parameters.

[0134] The update process employs a smooth transition method, where the new parameter = 0.7 × old parameter + 0.3 × calculated compensation coefficient, avoiding the impact of sudden parameter changes on system stability. The updated model parameters are applied in real-time to current distribution control, achieving dynamic optimization management of the battery pack's charging and discharging process.

[0135] In this embodiment of the invention, the detailed implementation steps of performing sliding window analysis on the charge migration rate and extracting the periodic fluctuation component of the charge migration rate under high-frequency charge-discharge excitation, denoted as the charge migration fluctuation characteristic, include:

[0136] The charge migration rate is segmented into sliding windows of a preset window length to obtain a subsequence of charge migration rate within each sliding window. Sliding window segmentation is a fundamental operation for discovering periodic patterns in time series data, revealing global regularities through local analysis. The segmentation process sets an appropriate window length (typically 3-5 times the high-frequency charge-discharge cycle) and slides along the time axis with a fixed step size (typically 10%-25% of the window length) to extract the data subsequence at each window position. The choice of window length considers the balance between the periodicity of the signal and computational complexity, ensuring both the capture of complete periodic information and efficient computation. Sliding window segmentation transforms long time series data into multiple short sequences that can be analyzed in parallel, laying the foundation for subsequent periodic analysis.

[0137] Autocorrelation analysis was performed on the charge migration rate subsequence to calculate the autocorrelation coefficient sequence. Autocorrelation analysis is an effective method for discovering the inherent periodicity of a signal, revealing repetitive patterns through the correlation between the signal and its delayed versions. The analysis process calculated the autocorrelation coefficients for each subsequence under different delays, forming an autocorrelation coefficient sequence. The standard autocorrelation function was used for the calculation.

[0138] ;

[0139] in, For delay The autocorrelation coefficient under the following conditions for The charge transfer rate at time t, For average speed, The length of the subsequence. for The charge transfer rate at time t.

[0140] The autocorrelation coefficient ranges from -1 to 1. A value closer to 1 indicates a higher similarity of signals at that delay, suggesting the possibility of periodicity. Autocorrelation analysis transforms the periodic characteristics of a time-domain signal into peak characteristics in the autocorrelation domain, facilitating accurate identification of the period length.

[0141] Peak points are extracted from the autocorrelation coefficient sequence, and the time interval between adjacent peak points is calculated, denoted as the fluctuation period. Peak extraction is a crucial step in determining the period length, achieved by locating local maxima in the autocorrelation sequence. A peak detection algorithm is employed, with an appropriate salience threshold (typically 0.3-0.5) set to ensure statistical significance of the identified peaks. The time interval between adjacent peaks directly corresponds to the signal period length; a stable and reliable fluctuation period estimate is obtained by averaging multiple intervals. The fluctuation period is the inherent rhythm of charge migration, reflecting the battery's response characteristics under high-frequency operating conditions, and provides a frequency constraint for the extraction of periodic fluctuation components.

[0142] The periodic fluctuation component of the charge migration rate is determined by the ratio of the fluctuation period to the high-frequency charge / discharge excitation frequency, and is denoted as the charge migration fluctuation characteristic. The high-frequency charge / discharge excitation frequency is the frequency of the fast charge / discharge signal actively applied to the battery pack by the system, rather than a value derived from battery characteristics. In other words, the high-frequency charge / discharge excitation frequency is the frequency of the periodic charge / discharge pulses applied to the battery by the power conversion device, typically in the range of several Hz to several kiloHz.

[0143] Frequency ratio analysis is a crucial step in correlating battery response with external excitation, identifying fluctuation components related to parasitic capacitance through frequency relationships. The analysis process calculates the ratio of the frequency corresponding to the fluctuation period to the charge / discharge excitation frequency, determining the type of frequency relationship (fundamental frequency, harmonic, or sub-frequency). Based on the frequency relationship, a corresponding bandpass filter is designed to extract the fluctuation components of the target frequency band from the original signal. The extracted result is the charge migration fluctuation characteristic, which directly reflects the battery's dynamic response behavior under high-frequency conditions, particularly the fluctuation patterns related to parasitic capacitance characteristics, providing a core basis for subsequent parasitic capacitance parameter identification.

[0144] In this embodiment of the invention, the detailed implementation steps for performing shortest path analysis on the local topological subgraph, identifying the shortest path connecting impedance abrupt changes, and marking the shortest path as the core influencing path include:

[0145] In the local topological subgraph, starting from the impedance abrupt change point, the path length to each adjacent node is calculated. The path length is determined by weighting the normalized impedance values ​​of the edges.

[0146] The normalized impedance value of an edge is actually the edge weight, which is calculated by the difference in the normalized impedance values ​​of adjacent battery cells.

[0147] Path length calculation is a fundamental step in shortest path analysis, defining the concept of "distance" between nodes through edge weights. The calculation process first identifies all directly adjacent nodes of abrupt changes, then determines the initial path length using edge weights. To amplify the impact of impedance differences, a weighted calculation method is used for path length:

[0148] ;

[0149] in, For the node To the node Path length, The weights of the connecting edges, The normalized impedance value of the target node. This is the weighting coefficient (usually ranging from 0.5 to 2).

[0150] This weighted calculation method considers not only the edge weights (impedance differences) but also the impedance characteristics of the target node, making path selection more reasonable and facilitating the identification of paths with significant impedance characteristics. The calculation results initialize the distance matrix of the shortest path algorithm, providing basic data for subsequent path searching.

[0151] Based on path length, Dijkstra's algorithm is used to calculate the shortest path from the impedance mutation point to all other nodes in the local topological subgraph. Dijkstra's algorithm is a classic method for solving single-source shortest paths, gradually determining the optimal path through a greedy strategy. The algorithm execution process includes four core steps: initialization, node selection, distance update, and path recording. The specific implementation considers the special characteristics of battery networks and makes appropriate optimizations to the algorithm. Optimizations include: using a priority queue to improve node selection efficiency; sparse graph representation to reduce storage overhead; and early termination conditions to avoid unnecessary computation. The algorithm outputs the shortest path from the mutation point to all other nodes and its length, providing a candidate set for subsequent path selection. The application of Dijkstra's algorithm ensures optimality and computational efficiency in path search, making it suitable for handling medium-sized graph structure problems such as battery networks.

[0152] The path with the smallest sum of edge weights among the shortest paths is extracted and designated as the candidate core path. The sum of edge weights is a comprehensive indicator of the "electrical distance" of a path, reflecting the cumulative effect of impedance changes along the path. The extraction process sorts all paths obtained by Dijkstra's algorithm and selects the path with the smallest sum of edge weights as a candidate. If multiple paths have similar sums of weights (difference less than 5%), they are retained as candidates for further verification. Candidate paths represent potential core influencing paths, signifying the most significant propagation channels of impedance changes, but their correlation with inconsistencies in parasitic capacitance needs to be confirmed through fluctuation verification.

[0153] Impedance fluctuation verification is performed on candidate core paths. The mean value of the local impedance fluctuation of each edge on the candidate core path is calculated, and candidate core paths with a mean value greater than a preset verification threshold are marked as core influence paths. Fluctuation verification is a key step in confirming the correlation between the path and parasitic capacitance inconsistency. The significance of the path's fluctuation is evaluated through statistical characteristics. The verification process calculates the mean value of the local impedance fluctuation of all edges on the path and compares it with a preset verification threshold. The preset verification threshold is usually set to 1.5-2 times the global average fluctuation to ensure that the identified paths have significant fluctuation characteristics. Candidate paths with a mean value greater than the threshold are finally confirmed as core influence paths. These paths are the main channels for the propagation of the parasitic capacitance inconsistency effect, directly affecting the balance of current distribution and providing key constraint information for subsequent coupling model construction.

[0154] In this embodiment of the invention, the detailed implementation steps for constructing a parasitic capacitance nonlinear coupling model based on the global topological characteristics of the current transient distribution interference factor and the parasitic capacitance weighted network model include:

[0155] The global clustering coefficient of the parasitic capacitance-weighted network model is calculated and denoted as the global topological feature value. The global clustering coefficient is a key indicator describing the overall connectivity of the network, reflecting the coupling strength within the battery pack. The calculation uses the standard network clustering coefficient formula, considering the connection density around each node:

[0156] ;

[0157] in, The global clustering coefficient is... The total number of network nodes. For nodes The number of connected edges. For nodes The degree (the number of nodes directly connected to it).

[0158] The global clustering coefficient ranges from [0,1]. A higher value indicates a tighter network connection and stronger coupling between battery cells; a lower value indicates a looser network connection and relatively independent battery cells. This metric provides a global reference for subsequent coupling strength adjustment, ensuring the model's adaptability and accuracy.

[0159] The interference propagation weights of each node in the parasitic capacitance-weighted network model are determined by multiplying the global topological eigenvalues ​​and the transient current allocation interference factor. Interference propagation weights are key indicators for quantifying the influence of nodes in unbalanced propagation, determining the weight of each battery cell in current allocation. The determination process combines global features with local interference factors to calculate the comprehensive influence of each node. Global eigenvalues ​​serve as basic coefficients, reflecting the overall network environment; local interference factors reflect the individual characteristics of nodes. Combining these two factors ensures both global consistency of the model and preserves the individual differences of nodes, making the weight allocation more reasonable and accurate. The interference propagation weights directly affect the subsequent construction of the adjacency matrix, determining the weight distribution of current allocation.

[0160] Based on interference propagation weights, a weighted adjacency matrix for a parasitic capacitance-weighted network model is constructed. The weighted adjacency matrix is ​​the mathematical expression of the network model, providing the data foundation for subsequent nonlinear transformations. The construction process is based on the original network topology, using interference propagation weights as node weights to update the adjacency matrix element values. The matrix element values ​​simultaneously consider node weights and edge weights, comprehensively reflecting the electrical coupling relationships between battery cells. The constructed adjacency matrix is ​​an N×N square matrix (N is the number of battery cells), with each element representing the coupling strength between the corresponding two nodes. The matrix exhibits good sparsity, facilitating efficient storage and computation, and providing structured input for subsequent nonlinear transformations of the coupled model.

[0161] A nonlinear transformation is performed on the weighted adjacency matrix to generate a parasitic capacitance nonlinear coupling model. This model determines the predicted transient current distribution and compensation requirements for each battery cell through eigenvalue decomposition of the weighted adjacency matrix. Eigenvalue decomposition is the core algorithm for constructing the coupling model, revealing the network's inherent dynamic characteristics through matrix decomposition. The decomposition process first performs eigenvalue decomposition on the adjacency matrix, then constructs a dynamic response model based on the main eigenvalues ​​and their corresponding eigenvectors. The eigenvectors directly map to the current distribution coefficients of the battery cells, while the eigenvalues ​​reflect the significance of different distribution patterns. The model construction considers the top K main eigenvalues ​​(typically 20%-30% of the total) to ensure that the model captures key dynamic characteristics while avoiding excessive complexity. The final coupling model includes the predicted current distribution and required compensation adjustment for each battery cell, providing accurate decision-making basis for charge and discharge control and achieving precise balanced control of current distribution within the battery pack.

[0162] The foregoing described a method for managing multiple batteries in parallel in an intelligent lithium battery pack according to an embodiment of this application. The following describes a management system for managing multiple batteries in parallel in an intelligent lithium battery pack according to an embodiment of this application. Please refer to [link to relevant documentation]. Figure 2 One embodiment of a smart lithium battery pack multi-cell parallel management system in this application includes:

[0163] The data acquisition module is used to collect transient charge distribution data and electrode interface impedance data of each battery cell in the parallel lithium battery pack during high-frequency charging and discharging.

[0164] The charge migration analysis module is used to calculate the charge migration rate of each battery cell during high-frequency charging and discharging based on transient charge distribution data, and to extract the dynamic offset characteristics of parasitic capacitance of each battery cell based on the charge migration rate.

[0165] The impedance network modeling module is used to construct the impedance distribution network model of each battery cell based on the electrode interface impedance data, and to identify the core influence path of parasitic capacitance inconsistency through local topological analysis of impedance abrupt change points in the impedance distribution network model.

[0166] The coupling model construction module is used to construct a nonlinear coupling model of parasitic capacitance based on the dynamic offset characteristics and core influence paths of parasitic capacitance during high-frequency charging and discharging.

[0167] The distortion feature extraction module is used to extract the distortion components related to the parasitic capacitance nonlinear coupling model in the charge pulse signal by real-time monitoring of the time-domain distortion characteristics of the charge pulse signal of each battery cell.

[0168] The dynamic adjustment module is used to dynamically adjust the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model based on the time-domain offset and amplitude attenuation rate of the distortion component.

[0169] The charge and discharge control module is used to control the charge and discharge management equipment of the parallel lithium battery pack based on the parameters of the adjusted parasitic capacitance nonlinear coupling model, so as to achieve precise balance of transient current distribution among the battery cells.

[0170] The modules are connected via wired and / or wireless means to enable data transmission between them.

[0171] This invention acquires real-time transient charge distribution data and electrode interface impedance data of each battery cell in a parallel lithium battery pack during high-frequency charging and discharging. It then constructs a precise model of the dynamic offset characteristics of parasitic capacitance and the impedance distribution network. Combined with a parasitic capacitance nonlinear coupling model for transient current distribution, it generates a current distribution scheme optimized for distortion characteristics. Through dynamic monitoring and parameter adjustment, it achieves precise balance in current distribution among battery cells. This highly adaptable approach optimizes charging and discharging parameters in real-time based on differences in battery cells and parasitic capacitance characteristics, significantly improving battery pack utilization efficiency and extending battery pack lifespan.

[0172] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0173] It should be noted that all formulas in this manual are calculated by removing dimensions and taking their numerical values. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0174] Although embodiments of the invention have been shown and described, those skilled in the art will understand 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 claims and their equivalents.

Claims

1. A smart lithium battery pack multi-cell parallel management system, characterized in that, include: The data acquisition module is used to collect transient charge distribution data and electrode interface impedance data of each battery cell in the parallel lithium battery pack during high-frequency charging and discharging. The charge migration analysis module is used to calculate the charge migration rate of each battery cell during high-frequency charging and discharging based on the transient charge distribution data, and to extract the parasitic capacitance dynamic shift characteristics of each battery cell based on the charge migration rate. The impedance network modeling module is used to construct an impedance distribution network model for each battery cell based on the electrode interface impedance data, and to identify the core influence path of parasitic capacitance inconsistency through local topological analysis of impedance abrupt change points in the impedance distribution network model. The coupling model construction module is used to construct a parasitic capacitance nonlinear coupling model for transient current distribution during high-frequency charging and discharging based on the parasitic capacitance dynamic offset characteristics and the core influence path. The distortion feature extraction module is used to extract the distortion components related to the parasitic capacitance nonlinear coupling model in the charge pulse signal by real-time monitoring of the time-domain distortion characteristics of the charge pulse signal of each battery cell. The dynamic adjustment module is used to dynamically adjust the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model according to the time-domain offset and amplitude attenuation rate of the distortion component. The charge / discharge control module is used to control the charge / discharge management equipment of the parallel lithium battery pack based on the parameters of the adjusted parasitic capacitance nonlinear coupling model, so as to achieve precise balance of transient current distribution among the battery cells.

2. The intelligent lithium battery pack multi-cell parallel management system according to claim 1, characterized in that, The step of calculating the charge migration rate of each battery cell during high-frequency charging and discharging based on the transient charge distribution data, and extracting the dynamic offset characteristics of the parasitic capacitance of each battery cell based on the charge migration rate, includes: The transient charge distribution data is segmented in the time domain to obtain the charge accumulation in each time domain segment, which is denoted as the segmented charge amount. Calculate the difference in charge between adjacent time domain segments, and in conjunction with the duration of the time domain segments, calculate the charge migration rate of each battery cell. A sliding window analysis was performed on the charge migration rate to extract the periodic fluctuation component of the charge migration rate under high-frequency charge and discharge excitation, which was denoted as the charge migration fluctuation feature. Based on the amplitude and phase changes of the charge migration fluctuation characteristics, and combined with a preset parasitic capacitance reference model, the parasitic capacitance dynamic offset characteristics of each battery cell are calculated. The parasitic capacitance dynamic offset characteristics are determined by the deviation between the amplitude of the charge migration fluctuation characteristics and the parasitic capacitance reference model.

3. The intelligent lithium battery pack multi-cell parallel management system according to claim 1, characterized in that, The process involves constructing an impedance distribution network model for each battery cell based on the electrode interface impedance data, and identifying the core influencing paths of parasitic capacitance inconsistencies through local topological analysis of impedance abrupt change points in the impedance distribution network model, including: The electrode interface impedance data is normalized to obtain the normalized impedance value of each battery cell. The impedance distribution network model is constructed by using each battery cell as a node and the difference in normalized impedance between adjacent battery cells as the edge weight. In the impedance distribution network model, the local standard deviation of the edge weight of each node is calculated and denoted as the local impedance variability. Nodes whose local impedance fluctuation is greater than a preset fluctuation threshold are marked as impedance abrupt change points, and the set of edges directly connected to these impedance abrupt change points is extracted and denoted as a local topological subgraph. Shortest path analysis is performed on the local topology subgraph to identify the shortest path connecting the impedance abrupt change point, and the shortest path is marked as the core influence path.

4. The intelligent lithium battery pack multi-cell parallel management system according to claim 1, characterized in that, The parasitic capacitance nonlinear coupling model for transient current distribution during high-frequency charging and discharging, based on the dynamic offset characteristics of the parasitic capacitance and the core influence path, includes: Based on the parasitic capacitance dynamic offset characteristics, the parasitic capacitance offset of each battery cell is calculated, and the node attributes of the impedance distribution network model are updated with the parasitic capacitance offset as the weight to generate a parasitic capacitance weighted network model. In the parasitic capacitance weighted network model, the set of edges corresponding to the core influence path is marked to generate a parasitic capacitance influence subgraph containing the core influence path; Based on the parasitic capacitance influence subgraph, the current transient distribution interference factor of each edge on the core influence path is calculated. The current transient distribution interference factor is determined by the nonlinear combination of the parasitic capacitance offset of the edge and the impedance value of the edge. Based on the global topological characteristics of the current transient distribution interference factor and the parasitic capacitance weighted network model, the parasitic capacitance nonlinear coupling model is constructed.

5. The intelligent lithium battery pack multi-cell parallel management system according to claim 1, characterized in that, The step of extracting the distortion components related to the parasitic capacitance nonlinear coupling model from the charge pulse signals by real-time monitoring of the time-domain distortion characteristics of the charge pulse signals of each battery cell includes: The charge pulse signals of each battery cell are collected, and the charge pulse signals are sampled in the time domain to obtain discrete time series data; Calculate the standard pulse template of the charge pulse signal, and compare the actual collected charge pulse signal with the standard pulse template to obtain the difference signal; The difference signal is subjected to threshold judgment. When the amplitude of the difference signal exceeds the preset distortion threshold, it is marked as a distortion time point. The time interval between adjacent distortion time points is statistically analyzed. When the deviation between the time interval and the characteristic period of the parasitic capacitance nonlinear coupling model is less than a preset deviation threshold, the corresponding difference signal segment is extracted as a distortion component.

6. The intelligent lithium battery pack multi-cell parallel management system according to claim 1, characterized in that, The step of dynamically adjusting the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model based on the time-domain offset and amplitude attenuation rate of the distortion component includes: Determine the start time of the distorted component and calculate its time difference with the start time of the ideal pulse signal to obtain the time domain offset; The peak amplitude and the end amplitude of the distortion component are measured, and the amplitude attenuation rate is calculated. The amplitude attenuation rate is equal to the difference between the peak amplitude and the end amplitude divided by the distortion duration. A first compensation coefficient is set based on the time domain offset. When the time domain offset is greater than a first threshold, the first compensation coefficient is increased; when the time domain offset is less than a second threshold, the first compensation coefficient is decreased. A second compensation coefficient is set based on the amplitude attenuation rate. When the amplitude attenuation rate is greater than the attenuation threshold, the adjustment range of the second compensation coefficient is increased. The first compensation coefficient and the second compensation coefficient are weighted and combined to obtain the final current distribution compensation coefficient, and the corresponding parameters of the parasitic capacitance nonlinear coupling model are updated.

7. The intelligent lithium battery pack multi-cell parallel management system according to claim 2, characterized in that, The process of performing a sliding window analysis on the charge migration rate to extract the periodic fluctuation component of the charge migration rate under high-frequency charge-discharge excitation, denoted as the charge migration fluctuation feature, includes: The charge migration rate is divided into sliding windows with a preset window length to obtain a subsequence of charge migration rate within each sliding window; Autocorrelation analysis is performed on the charge mobility rate subsequence to calculate the autocorrelation coefficient sequence of the charge mobility rate subsequence; Peak points are extracted from the autocorrelation coefficient sequence, and the time interval between adjacent peak points is calculated and denoted as the fluctuation period; The periodic fluctuation component of the charge migration rate is determined based on the ratio of the fluctuation period to the high-frequency charge-discharge excitation frequency, and is denoted as the charge migration fluctuation characteristic.

8. The intelligent lithium battery pack multi-cell parallel management system according to claim 3, characterized in that, The step of performing shortest path analysis on the local topology subgraph to identify the shortest path connecting the impedance abrupt change point and marking the shortest path as the core influencing path includes: In the local topological subgraph, starting from the impedance abrupt change point, the path length to each adjacent node is calculated, and the path length is determined by weighting the normalized impedance values ​​of the edges. Based on the path length, Dijkstra's algorithm is used to calculate the shortest path from the impedance abrupt change point to all other nodes in the local topology subgraph; Extract the path with the smallest sum of edge weights from the shortest paths, and denote it as the candidate core path; Impedance fluctuation verification is performed on the candidate core path. The mean value of the local impedance fluctuation of each side on the candidate core path is calculated. The candidate core path with the mean value greater than the preset verification threshold is marked as the core influence path.

9. The intelligent lithium battery pack multi-cell parallel management system according to claim 4, characterized in that, The parasitic capacitance nonlinear coupling model is constructed based on the global topological features of the current transient distribution interference factor and the parasitic capacitance weighted network model, including: Calculate the global clustering coefficients of the parasitic capacitance weighted network model, and denote them as global topological feature values; The interference propagation weight of each node in the parasitic capacitance weighted network model is determined by multiplying the global topological feature value with the current transient allocation interference factor. Based on the interference propagation weights, the weighted adjacency matrix of the parasitic capacitance weighted network model is constructed; The weighted adjacency matrix is ​​subjected to a nonlinear transformation to generate the parasitic capacitance nonlinear coupling model. The parasitic capacitance nonlinear coupling model determines the predicted value of the transient current distribution and the compensation requirement value of each battery cell through the eigenvalue decomposition of the weighted adjacency matrix.

10. A method for managing multiple parallel cells in an intelligent lithium battery pack, implemented based on the intelligent lithium battery pack multi-cell parallel management system according to any one of claims 1 to 9, characterized in that, include: Step 1: Collect transient charge distribution data and electrode interface impedance data of each battery cell in the parallel lithium battery pack during high-frequency charging and discharging. Step 2: Based on the transient charge distribution data, calculate the charge migration rate of each battery cell during the high-frequency charging and discharging process, and extract the parasitic capacitance dynamic shift characteristics of each battery cell based on the charge migration rate. Step 3: Based on the electrode interface impedance data, construct an impedance distribution network model for each battery cell, and identify the core influencing path of parasitic capacitance inconsistency through local topological analysis of impedance abrupt change points in the impedance distribution network model. Step 4: Based on the dynamic offset characteristics of the parasitic capacitance and the core influence path, construct a nonlinear coupling model of parasitic capacitance for transient current distribution during high-frequency charging and discharging. Step 5: By real-time monitoring of the time-domain distortion characteristics of the charge pulse signals of each battery cell, extract the distortion components in the charge pulse signals that are related to the nonlinear coupling model of the parasitic capacitance; Step 6: Dynamically adjust the current distribution compensation coefficient in the parasitic capacitance nonlinear coupling model based on the time-domain offset and amplitude attenuation rate of the distortion component. Step 7: Based on the adjusted parameters of the parasitic capacitance nonlinear coupling model, control the charging and discharging management device of the parallel lithium battery pack.

Images