Battery performance testing method and system

Abstract

The present application relates to the technical field of battery performance testing, and discloses a battery performance testing method and system. The method includes: obtaining charging performance data of multiple battery cells in a target battery pack, and constructing a current balancing path network of multiple battery cells based on the charging performance data; performing network feature calculation on the current balancing path network, obtaining a network feature set and performing performance test strategy analysis, obtaining a charge and discharge cycle strategy and a temperature disturbance strategy; performing a charge and discharge cycle test based on the charge and discharge cycle strategy, obtaining charge and discharge cycle test data; performing a temperature disturbance and recovery test based on the temperature disturbance strategy, obtaining temperature disturbance test data; performing current balancing control analysis on the charge and discharge cycle test data and the temperature disturbance test data through the MPC‑LQG model, obtaining current balancing control parameters, and the present application improves the accuracy of the battery performance test and thus improves the performance of the battery.

Description

Technical Field

[0001] The present application relates to the technical field of battery performance testing, and in particular to a battery performance testing method and system. Background Art

[0002] The stability and reliability of battery performance face huge challenges. During long-term use, battery cells will experience performance degradation, increased internal resistance, and capacity attenuation. These problems not only affect the service life of the equipment, but may also cause safety hazards.

[0003] Existing battery performance testing methods mainly focus on testing single factors, such as charge and discharge performance, temperature response, etc., and lack comprehensive and systematic testing methods. These methods often cannot accurately reflect the true performance of the battery under complex usage environments. For example, in actual applications, the individual battery cells inside the battery pack may exhibit different charge and discharge characteristics and temperature responses due to differences in manufacturing processes and usage conditions. These differences will lead to uneven current distribution, which in turn affects the performance and life of the entire battery pack. Traditional testing methods cannot fully capture these dynamic changes and interactions, thus limiting their effectiveness in practical applications. In addition, existing methods also have shortcomings in handling data analysis and model prediction. Most methods rely only on simple linear models and cannot accurately describe the nonlinear characteristics and complex dynamic behaviors of battery performance. Summary of the invention

[0004] The present application provides a battery performance testing method and system, which are used to improve the accuracy of battery performance testing and thus improve battery performance.

[0005] In a first aspect, the present application provides a battery performance testing method, the battery performance testing method comprising:

[0006] Acquire charging performance data of a plurality of battery cells in a target battery pack, and construct a current balancing path network of the plurality of battery cells according to the charging performance data;

[0007] Performing network feature calculation on the current balancing path network to obtain a network feature set, and performing performance test strategy analysis on the target battery pack according to the network feature set to obtain a charge-discharge cycle strategy and a temperature disturbance strategy;

[0008] Performing a charge-discharge cycle test on the target battery pack based on the charge-discharge cycle strategy to obtain charge-discharge cycle test data;

[0009] Based on the temperature disturbance strategy, performing temperature disturbance and recovery tests on the target battery pack to obtain temperature disturbance test data;

[0010] The MPC-LQG model is used to perform current balancing control analysis on the charge-discharge cycle test data and the temperature disturbance test data to obtain current balancing control parameters.

[0011] In a second aspect, the present application provides a battery performance testing system, the battery performance testing system comprising:

[0012] An acquisition module, used to acquire charging performance data of a plurality of battery cells in a target battery pack, and to construct a current balancing path network of the plurality of battery cells according to the charging performance data;

[0013] A calculation module, used to perform network feature calculation on the current balancing path network to obtain a network feature set, and perform performance test strategy analysis on the target battery pack according to the network feature set to obtain a charge-discharge cycle strategy and a temperature disturbance strategy;

[0014] A cycle test module, used to perform a charge and discharge cycle test on the target battery pack based on the charge and discharge cycle strategy to obtain charge and discharge cycle test data;

[0015] A disturbance test module, used to perform temperature disturbance and recovery tests on the target battery pack based on the temperature disturbance strategy to obtain temperature disturbance test data;

[0016] The analysis module is used to perform current balancing control analysis on the charge-discharge cycle test data and the temperature disturbance test data through the MPC-LQG model to obtain current balancing control parameters.

[0017] In the technical solution provided by the present application, the charging performance data of multiple battery cells in the target battery pack are obtained, and a current balancing path network is constructed based on these data, so as to realize accurate identification and positioning of the battery cells. This step not only covers the comprehensive measurement and normalization of key parameters such as the internal resistance, SOC (state of charge) and SOH (health state) of the battery cells, but also effectively solves the problem of uneven performance between battery cells by constructing a current balancing path network. It can ensure that the current distribution of each unit in the entire battery pack is uniform during the charging and discharging process, and improve the overall performance and life of the battery pack. The present invention calculates the network characteristics of the current balancing path network in detail, generates a network feature set, and analyzes the performance test strategy based on these characteristics to obtain the charge and discharge cycle strategy and the temperature perturbation strategy. Through the comprehensive analysis of multi-dimensional features such as node degree, path length, clustering coefficient, same-layer degree and inter-layer degree, the accurate evaluation and prediction of battery pack performance is realized. This method can not only identify potential performance bottlenecks in the battery pack, but also formulate optimized charge and discharge and temperature perturbation strategies, so as to better manage the use and maintenance of the battery pack. By introducing the MPC-LQG model, the current balancing control analysis is performed on the charge and discharge cycle test data and the temperature disturbance test data to obtain the current balancing control parameters. The MPC-LQG model combines the advantages of model predictive control (MPC) and linear quadratic Gaussian (LQG) control, and can adjust the control strategy in real time under dynamic environments. The current distribution and charging process are further optimized through curve fitting, feature recognition and state space model construction. This not only improves the stability and reliability of the battery pack under different working conditions, but also effectively extends the battery life, thereby improving the performance of the battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0019] Figure 1 A schematic diagram of an embodiment of a battery performance testing method in an embodiment of the present application;

[0020] Figure 2 This is a schematic diagram of an embodiment of a battery performance testing system in an embodiment of the present application. DETAILED DESCRIPTION

[0021] The present application embodiment provides a battery performance test method and system. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the battery performance testing method in the embodiment of the present application includes:

[0023] Step S101, acquiring charging performance data of a plurality of battery cells in a target battery pack, and constructing a current balancing path network of the plurality of battery cells according to the charging performance data;

[0024] It is understandable that the execution subject of the present application may be a battery performance test system, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0025] Specifically, the target battery pack is identified and located by battery cell identification and positioning, and multiple battery cells and the spatial position information of each battery cell are obtained. The historical data of multiple battery cells are collected to obtain the internal resistance data, SOC (State of Charge) data and SOH (State of Health) data of each battery cell, which reflect the historical charging performance and current state of each battery cell. The internal resistance data, SOC data and SOH data of each battery cell are scaled and normalized to obtain the charging performance data of each battery cell. Data of different dimensions are converted to the same dimension to ensure the accuracy and consistency of subsequent calculations and analyses. Nodes of multiple battery cells are initialized according to the spatial position information to obtain the initial node of each battery cell. By combining the spatial position information and the charging performance data, a unique initial node is assigned to each battery cell. The initial current balancing path of each initial node is created according to the charging performance data, reflecting the potential path for the balanced distribution of current between different battery cells. The path weights are calculated and the nodes are assigned for the initial current balancing paths respectively to obtain multiple target current balancing paths and multiple target nodes. The path weight calculation takes into account the charging performance data of each battery cell and determines the weight of each path based on this data, thereby reflecting the flow resistance and priority of the current in the path. Node assignment assigns corresponding attributes and values ​​to each target node, completing the structure of the current balancing path. A current balancing path network of multiple battery cells is constructed based on multiple target current balancing paths and multiple target nodes. All target current balancing paths and target nodes are combined to form a complete current balancing path network.

[0026] Step S102, calculating network characteristics of the current balancing path network to obtain a network characteristic set, and performing performance test strategy analysis on the target battery pack according to the network characteristic set to obtain a charge-discharge cycle strategy and a temperature disturbance strategy;

[0027] Specifically, the node degree of the current balancing path network is calculated to obtain the node degree. The node degree reflects the degree of connection between each node and other nodes, and is a key indicator for evaluating the importance and influence of the battery cell in the entire network. At the same time, the path length of the current balancing path network is calculated to obtain the path length. The path length represents the distance of the shortest path between two nodes, which helps to understand the transmission efficiency of current between battery cells. The clustering coefficient of the current balancing path network is calculated to obtain the clustering coefficient. The clustering coefficient is used to measure the closeness between the neighbors of the node and reflect the local clustering of the network. The same-layer degree of the current balancing path network is calculated to obtain the same-layer degree. The same-layer degree refers to the connection strength between nodes at the same level, which is an important parameter for understanding the interaction between nodes at the same level. At the same time, the inter-layer degree of the current balancing path network is calculated to obtain the inter-layer degree. The inter-layer degree measures the connection strength between nodes at different levels, which helps to analyze the cross-level current balancing path. According to the calculated node degree, path length, clustering coefficient, same-layer degree and inter-layer degree, the corresponding network feature set is generated. The feature set integrates various important parameters in the network and comprehensively describes the structure and performance of the current balancing path network. The network feature set and multiple preset candidate cycle strategies are correlated to obtain the first correlation of each candidate cycle strategy. Through correlation calculation, determine which charge and discharge cycle strategies best match the network feature set, and select the charge and discharge cycle strategy that best matches the target battery pack. At the same time, the network feature set and multiple preset candidate perturbation strategies are correlated to obtain the second correlation of each candidate perturbation strategy. Through correlation calculation, evaluate the matching degree of different temperature perturbation strategies with the network feature set, and select the temperature perturbation strategy that best matches the target battery pack.

[0028] Step S103, performing a charge-discharge cycle test on the target battery pack based on the charge-discharge cycle strategy to obtain charge-discharge cycle test data;

[0029] Specifically, based on the charge and discharge cycle strategy, multiple charging tests are performed on the target battery pack to obtain charging process data, and at the same time, multiple discharge tests are performed on the target battery pack to obtain discharge process data. The charging process data is linearly transformed to obtain a charging linear feature set, and the discharge process data is linearly transformed to obtain a discharge linear feature set. Ensuring the comparability and consistency of the data is helpful for the accuracy of subsequent analysis. The charging covariance matrix elements are calculated according to the multiple charging linear features in the charging linear feature set, and the charging covariance matrix is ​​constructed according to the charging covariance matrix elements; similarly, the discharge covariance matrix elements are calculated according to the multiple discharge linear features in the discharge linear feature set, and the discharge covariance matrix is ​​constructed according to the discharge covariance matrix elements. On this basis, the correlation between the multiple charging linear features is calculated according to the charging covariance matrix to obtain the charging feature correlation; similarly, the correlation between the multiple discharge linear features is calculated according to the discharge covariance matrix to obtain the discharge feature correlation. Through correlation analysis, the intrinsic connection and mutual influence between the various features in the charging and discharging process are identified. According to the correlation of charging characteristics, multiple charging characteristic values ​​corresponding to multiple charging linear characteristics are calculated; according to the correlation of discharging characteristics, multiple discharging characteristic values ​​corresponding to multiple discharging linear characteristics are calculated. The characteristic value reflects the importance and contribution of each characteristic in the entire charging and discharging process. According to multiple charging characteristic values, the corresponding charging characteristic vector is generated, and the charging characteristic vector is normalized by the principal component to obtain the charging principal component characteristic weight; similarly, according to multiple discharging characteristic values, the corresponding discharging characteristic vector is generated, and the discharging characteristic vector is normalized by the principal component to obtain the discharging principal component characteristic weight. The charging process data is multiplied by the charging principal component characteristic weight to obtain multiple charging products, and the multiple charging products are summed to obtain the charging characteristic data; similarly, the discharging process data is multiplied by the discharging principal component characteristic weight to obtain multiple discharging products, and the multiple discharging products are summed to obtain the discharging characteristic data. The key characteristic data in the charging and discharging process are extracted to fully reflect the performance of the battery under different charging and discharging conditions. The charging characteristic data and the discharging characteristic data are fused to obtain the charging and discharging cycle test data. Through data fusion, the important information in the charging and discharging process is integrated to form a complete performance test data set.

[0030] Step S104: Based on the temperature disturbance strategy, perform temperature disturbance and recovery tests on the target battery pack to obtain temperature disturbance test data;

[0031] Specifically, the temperature disturbance data of the target battery pack is set based on the temperature disturbance strategy. By setting a series of temperature change parameters, the battery pack works under specific temperature conditions to simulate the temperature fluctuations in the actual operating environment. The temperature disturbance data is subjected to disturbance gradient calculation to obtain the temperature disturbance gradient parameters. The disturbance gradient calculation determines the specific parameters of the temperature change at each stage by analyzing the rate and amplitude of the temperature change. These parameters are used to guide the subsequent charge and discharge disturbance test. According to the temperature disturbance gradient parameters, the target battery pack is subjected to charge and discharge disturbance test to obtain the recovery process data. The charge and discharge disturbance test performs charge and discharge operations on the battery under different temperature conditions to observe the performance changes and recovery process of the battery, thereby evaluating the adaptability of the battery under temperature fluctuation conditions. The temperature disturbance data and the recovery process data are aligned to obtain the initial disturbance test data. Each data point in the temperature disturbance process is matched with the data in the recovery process to ensure the accuracy and consistency of the analysis. The data at different stages are associated through timestamps or other tags to form a continuous test data set. The initial disturbance test data is subjected to threshold filtering to obtain the temperature disturbance test data. By setting a certain threshold, the abnormal data beyond the normal range in the initial disturbance test data is filtered out, ensuring that the final temperature disturbance test data is more accurate and reliable.

[0032] Step S105: Perform current balancing control analysis on the charge-discharge cycle test data and the temperature disturbance test data through the MPC-LQG model to obtain current balancing control parameters.

[0033] Specifically, curve fitting is performed on the charge-discharge cycle test data and the temperature disturbance test data to obtain the charge-discharge cycle test curve and the temperature disturbance test curve. The test data is described by a mathematical model to make it smoother and more continuous, so as to facilitate subsequent analysis and processing. Curve feature recognition is performed on the charge-discharge cycle test curve and the temperature disturbance test curve to obtain the charge-discharge cycle feature point set and the temperature disturbance feature point set. Feature recognition is to extract the key points of the curve and find the point set that can represent the overall characteristics of the curve. Feature mapping and vector conversion are performed on the charge-discharge cycle feature point set and the temperature disturbance feature point set to obtain the charge-discharge cycle feature vector and the temperature disturbance feature vector. Feature mapping and vector conversion are to convert the feature point set into a vector form so that it can be used in the state space model. The first state space model of the charge-discharge cycle feature vector and the second state space model of the temperature disturbance feature vector are defined. The first state space model includes a first state equation and a first output equation, and the second state space model includes a second state equation and a second output equation. The state space model describes the state and output of the system through mathematical equations so that it can be analyzed and optimized through control theory. A first prediction model is created based on the first state space model, and a second prediction model is created based on the second state space model, and the optimization objectives, optimization problems and constraints of the first prediction model and the second prediction model are defined. The prediction model predicts the future state based on the current state and historical data, and solves the optimal control strategy through the optimization algorithm. The constraints are used to limit the solution space in the optimization process so that the solution meets the actual physical constraints and system requirements. The LQR controllers of the first prediction model and the second prediction model are set respectively to obtain the MPC-LQG model. The LQR controller controls and optimizes the prediction model through the linear quadratic regulator method so that the system state can reach the expected goal. The MPC-LQG model is a method that combines model predictive control (Model Predictive Control) and linear quadratic Gaussian (Linear Quadratic Gaussian) control. The control strategy is optimized by comprehensive processing of the prediction model and noise. The current balancing control analysis of the charge and discharge cycle characteristic vector and the temperature disturbance characteristic vector is performed through the MPC-LQG model to obtain the current balancing control parameters. The current distribution during the charge and discharge process and temperature disturbance process is precisely controlled to achieve current balance in the battery pack and improve the overall performance and service life of the battery.

[0034] The first prediction model is used to predict the current balancing control of the charge and discharge cycle characteristic vector to obtain the first initial control sequence. The prediction model is used to predict the optimal control strategy for current balancing in the future period of time according to the charge and discharge cycle characteristic vector, and the first initial control sequence is generated. At the same time, the second prediction model is used to predict the current balancing control of the temperature disturbance characteristic vector to obtain the second initial control sequence. According to the temperature disturbance characteristic vector, the prediction model is used to infer the optimal current balancing control strategy under temperature change conditions to form the second initial control sequence. The first initial control sequence is feedback-corrected by the LQR controller to obtain the first target control sequence. The LQR controller optimizes and adjusts the initial control sequence by minimizing a predefined performance indicator function to make it closer to the optimal control requirements of the actual system. Similarly, the LQR controller is feedback-corrected by the second initial control sequence to obtain the second target control sequence. The first target control sequence and the second target control sequence are weighted and parameter-mapped to obtain the current balancing control parameters. According to the different sensitivities of the system to the charge and discharge cycle and temperature disturbance, the two groups of control sequences are appropriately weighted so that the final control parameters can not only effectively balance the current, but also adapt to the influence of temperature changes. Parameter mapping converts the weighted fusion control strategy into specific control parameters for practical application.

[0035] In the embodiment of the present application, the charging performance data of multiple battery cells in the target battery pack are obtained, and a current balancing path network is constructed based on these data to achieve accurate identification and positioning of the battery cells. This step not only covers the comprehensive measurement and normalization of key parameters such as the internal resistance, SOC (state of charge) and SOH (health state) of the battery cells, but also effectively solves the problem of uneven performance between battery cells by constructing a current balancing path network. It can ensure that the current distribution of each unit in the entire battery pack is uniform during the charging and discharging process, thereby improving the overall performance and life of the battery pack. The present invention generates a network feature set by performing detailed network feature calculations on the current balancing path network, and analyzes the performance test strategy based on these features to obtain a charge and discharge cycle strategy and a temperature perturbation strategy. Through the comprehensive analysis of multi-dimensional features such as node degree, path length, clustering coefficient, same-layer degree and inter-layer degree, accurate evaluation and prediction of battery pack performance are achieved. This method can not only identify potential performance bottlenecks in battery packs, but also formulate optimized charge and discharge and temperature perturbation strategies to better manage the use and maintenance of battery packs. By introducing the MPC-LQG model, the current balancing control analysis is performed on the charge and discharge cycle test data and the temperature disturbance test data to obtain the current balancing control parameters. The MPC-LQG model combines the advantages of model predictive control (MPC) and linear quadratic Gaussian (LQG) control, and can adjust the control strategy in real time under dynamic environments. The current distribution and charging process are further optimized through curve fitting, feature recognition and state space model construction. This not only improves the stability and reliability of the battery pack under different working conditions, but also effectively extends the battery life, thereby improving the performance of the battery.

[0036] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0037] (1) Identify and locate the battery cells of the target battery pack to obtain multiple battery cells and spatial position information of each battery cell;

[0038] (2) Collect historical data of multiple battery cells to obtain internal resistance data, SOC data, and SOH data of each battery cell;

[0039] (3) performing scale normalization processing on the internal resistance data, SOC data, and SOH data of each battery cell to obtain the charging performance data of each battery cell;

[0040] (4) Initializing nodes of multiple battery cells according to spatial position information to obtain an initial node of each battery cell;

[0041] (5) creating an initial current balancing path for each initial node according to the charging performance data;

[0042] (6) performing path weight calculation and node assignment on the initial current balancing paths respectively to obtain multiple target current balancing paths and multiple target nodes;

[0043] (7) Constructing a current balancing path network of multiple battery cells according to multiple target current balancing paths and multiple target nodes.

[0044] Specifically, the battery pack is scanned and analyzed using image recognition technology or sensor data. Through image recognition, the location, number and relative position of each battery cell are identified. Sensor data can provide more detailed spatial location information, such as the precise coordinates of each battery cell. After obtaining this information, a battery pack spatial layout diagram is constructed to ensure that the location information of each battery cell is accurate. Historical data is collected for multiple battery cells to collect the internal resistance data, SOC (State of Charge) data and SOH (State of Health) data of each battery cell. The internal resistance data can be obtained by applying an AC signal to the battery and measuring its impedance. The SOC data is estimated by measuring the battery voltage and current and using the coulomb counting method. The SOH data needs to be calculated comprehensively based on multiple factors such as internal resistance and capacity attenuation. The internal resistance data, SOC data and SOH data of each battery cell are scaled and normalized to eliminate the differences between data of different dimensions and make them comparable, and the normalized internal resistance data, SOC data and SOH data, i.e., the charging performance data of each battery cell, are obtained. Based on the normalized charging performance data and spatial position information, nodes of multiple battery cells are initialized to obtain the initial nodes of each battery cell. An initial current balancing path is created for each initial node according to the charging performance data. The initial current balancing path refers to the flow path of the current between different nodes determined based on the charging performance data of each node. Assuming that the weight of the current balancing path is determined by the distance between nodes and the charging performance data, the path weight w can be defined. ij for:

[0045]

[0046] Among them, dist(N i ,N j ) is the Euclidean distance between nodes, is the difference between the charging performance data, and α and β are weight coefficients. The path weight calculation and node assignment are performed on the initial current balancing path respectively to obtain multiple target current balancing paths and multiple target nodes. The path weight calculation is performed according to the above formula, and the node assignment is to assign the charging performance data and position data of each node to the path weight model to form a complete path network. Through calculation, multiple target current balancing paths and multiple target nodes are obtained. Finally, according to the target current balancing paths and target nodes, a current balancing path network of multiple battery cells is constructed. The network is a graph structure, in which nodes represent battery cells, edges represent current balancing paths, and the weights of edges represent the priority or difficulty of current balancing. The network reflects the distribution and balancing of current in the battery pack, and provides a basis for optimizing current distribution for the battery management system. For example, suppose there is a battery pack consisting of four battery cells, and its spatial position information is: N 1 =(0,0,0,0.5,0.7,0.8),N 2 =(1,0,0,0.6,0.8,0.7),N 3 =(0,1,0,0.4,0.9,0.6),N 4 =(1,1,0,0.7,0.6,0.9). Calculate the path weights between nodes, for example, Assume that α=1 and β=1, and dist(N 1 ,N 2 )=1, Then 12 =1·1+1·0.2=1.2. Through this calculation method, the path weights between all nodes are obtained, thereby constructing a complete current balancing path network.

[0047] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0048] (1) Calculating the node degree of the current balancing path network to obtain the node degree, and calculating the path length of the current balancing path network to obtain the path length;

[0049] (2) calculating a clustering coefficient of the current balancing path network to obtain a clustering coefficient, calculating a same-layer degree of the current balancing path network to obtain a same-layer degree, and calculating an inter-layer degree of the current balancing path network to obtain an inter-layer degree;

[0050] (3) Generate a corresponding network feature set based on node degree, path length, clustering coefficient, same-layer degree, and inter-layer degree;

[0051] (4) performing correlation calculation on the network feature set and a plurality of preset candidate cycle strategies to obtain a first correlation of each candidate cycle strategy, and selecting a charge and discharge cycle strategy of the target battery pack according to the first correlation;

[0052] (5) performing correlation calculation on the network feature set and multiple preset candidate disturbance strategies to obtain a second correlation of each candidate disturbance strategy, and selecting a temperature disturbance strategy for the target battery pack according to the second correlation.

[0053] Specifically, the node degree of the current balancing path network is calculated to obtain the node degree. The node degree indicates the number of connections of each node, that is, how many paths are connected to the node. Assume that the current balancing path network can be represented as a graph G = (V, E), where V represents the node set and E represents the edge set. For each node v i , its degree d(v i ) is calculated as:

[0054]

[0055] Among them, d(v i ) represents node v i The degree of n is the total number of nodes, a ij is an element of the adjacency matrix, representing the node v i and node v j Is there a path connection between them? If so, then a ij =1, otherwise a ij =0. Perform path length calculation on the current balancing path network to obtain the path length. Path length refers to the distance of the shortest path between two nodes, calculated using breadth-first search (BFS) or Dijkstra algorithm. Assume l(v i ,v j ) represents node v i and node v j The shortest path length between, then the path length L can be expressed by the following formula:

[0056]

[0057] Where L represents the average path length of the entire network, represents the total number of nodes, l(v i ,v j ) is the node v i and node v j The shortest path length between i≠j It means that all node pairs with i≠j are summed. The clustering coefficient is calculated for the current balancing path network to obtain the clustering coefficient. The clustering coefficient measures the closeness between the neighbors of the node. Assume that the node vi The neighbor set of i ), then node v i The clustering coefficient C(v i ) is calculated as:

[0058]

[0059] Among them, C(v i ) represents node v i The clustering coefficient, {e jk} indicates that in the neighbor set N(v i ) in the edge set, indicating that at node v i Neighbor node v j and v k The actual number of edges between i ) represents node v i The same-level degree is calculated for the current balancing path network to obtain the same-level degree. The same-level degree indicates the connection strength between nodes at the same level. In a hierarchical network, assume that the node set V is divided into multiple levels V 1 ,V 2 ,...,V h , then the same level degree L s The calculation formula is:

[0060]

[0061] Among them, L s Indicates the same level, Indicates the same level V k The number of connections between nodes, is the total number of node pairs in all levels, Indicates level V k The total number of node pairs in |V k |It is level V k The number of nodes in the network. The inter-layer degree is calculated for the current balancing path network to obtain the inter-layer degree. The inter-layer degree measures the connection strength between nodes at different levels. Assume that V k and V l is a set of nodes at different levels, and the inter-level degree L ij The calculation formula is:

[0062]

[0063] Among them, L ij Indicates the interlayer degree, Indicates level V k and level V l The number of connections between |V k and |Vl| are the levels V k and level Vl The number of nodes in . Generate the corresponding network feature set according to the node degree, path length, clustering coefficient, same-layer degree and inter-layer degree. The feature set comprehensively reflects the structure and properties of the current balancing path network. The network feature set and multiple preset candidate cycle strategies are correlated to obtain the first correlation of each candidate cycle strategy, and the charge and discharge cycle strategy of the target battery pack is selected according to the first correlation. The correlation calculation can use the Pearson correlation coefficient or other correlation indicators. Assuming that the network feature set is F and the candidate cycle strategy set is S, the calculation formula of the correlation ρ(F,S) is:

[0064]

[0065] Among them, ρ(F,S) represents the correlation between the network feature set F and the candidate cycle strategy set S, F i and S i are the elements in the network feature set and the candidate cycle strategy set, respectively. and are their means respectively, and n is the number of features and strategies. The correlation between the network feature set and multiple preset candidate perturbation strategies is calculated to obtain the second correlation of each candidate perturbation strategy, and the temperature perturbation strategy of the target battery pack is selected according to the second correlation. The correlation calculation method is the same as above, ensuring that the selected temperature perturbation strategy has the best match with the network feature set, thereby optimizing the temperature response capability of the battery pack. For example, suppose there is a battery pack whose current balancing path network has the following characteristics: node degree is 3, path length is 5, clustering coefficient is 0.6, same-layer degree is 0.4, and inter-layer degree is 0.2. Based on these characteristics, a network feature set F = {3,5,0.6,0.4,0.2} is generated. In the preset candidate cycle strategy set, there are two strategies S 1 and S 2 , the corresponding feature set is S 1 ={2,5,0.7,0.3,0.1} and S 2 ={3,6,0.5,0.4,0.2}. Through correlation calculation, we get ρ(F,S 1 )=0.85 and ρ(F,S 2 )=0.95, so we choose the strategy S with higher correlation 2 As the charge and discharge cycle strategy of the target battery pack. For the temperature disturbance strategy, suppose there are two candidate strategies T 1 and T 2 , whose feature sets are T 1 ={3,4,0.6,0.5,0.3} and T 2 ={4,5,0.7,0.4,0.2}. Through correlation calculation, we get ρ(F,T1 )=0.88 and ρ(F,T 2 )=0.92, so we choose the strategy T with higher correlation. 2 As a temperature perturbation strategy for the target battery pack.

[0066] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0067] (1) Based on the charge and discharge cycle strategy, multiple charging tests are performed on the target battery pack to obtain charging process data, and multiple discharge tests are performed on the target battery pack to obtain discharge process data;

[0068] (2) performing linear transformation on the charging process data to obtain a charging linear feature set, and performing linear transformation on the discharging process data to obtain a discharging linear feature set;

[0069] (3) calculating charging covariance matrix elements according to multiple charging linear features in the charging linear feature set, and constructing a charging covariance matrix according to the charging covariance matrix elements; calculating discharging covariance matrix elements according to multiple discharging linear features in the discharging linear feature set, and constructing a discharging covariance matrix according to the discharging covariance matrix elements;

[0070] (4) calculating the correlation between multiple charging linear features according to the charging covariance matrix to obtain the charging feature correlation; calculating the correlation between multiple discharging linear features according to the discharging covariance matrix to obtain the discharging feature correlation;

[0071] (5) calculating a plurality of charging characteristic values ​​corresponding to a plurality of charging linear characteristics according to the charging characteristic correlation; and calculating a plurality of discharging characteristic values ​​corresponding to a plurality of discharging linear characteristics according to the discharging characteristic correlation;

[0072] (6) generating a corresponding charging feature vector according to the plurality of charging feature values, and performing principal component normalization on the charging feature vector to obtain a charging principal component feature weight; generating a corresponding discharging feature vector according to the plurality of discharging feature values, and performing principal component normalization on the discharging feature vector to obtain a discharging principal component feature weight;

[0073] (7) multiplying the charging process data by the charging principal component characteristic weight to obtain a plurality of charging products, and summing the plurality of charging products to obtain charging characteristic data; multiplying the discharging process data by the discharging principal component characteristic weight to obtain a plurality of discharging products, and summing the plurality of discharging products to obtain discharging characteristic data;

[0074] (8) Perform data fusion on the charging characteristic data and the discharging characteristic data to obtain the charging and discharging cycle test data.

[0075] Specifically, a specific charge and discharge cycle strategy is set. This includes setting parameters such as charging current, charging voltage, and charging time, and performing multiple charging tests on the battery pack under these parameters. Through the test, the voltage, current, temperature and other data of the battery during the charging process are recorded to form a complete charging process data set. Multiple discharge tests are performed on the target battery pack, and the voltage, current, temperature and other data generated during the discharge process are recorded by setting parameters such as discharge current, discharge voltage, and discharge time to form a discharge process data set. The charging process data is linearly transformed to convert data of different dimensions to the same dimension. The linear transformation can be performed by normalization or standardization methods, and a charging linear feature set is obtained through linear transformation. The same linear transformation is performed on the discharge process data to obtain a discharge linear feature set. The charging covariance matrix elements are calculated based on multiple charging linear features in the charging linear feature set, and the charging covariance matrix is ​​constructed based on these elements. Assume that the charging linear feature set is X, and its elements are X i , the covariance matrix element C ij The calculation formula is:

[0076]

[0077] Where n is the number of samples, X ik and X jk are the values ​​of the i-th and j-th features in the k-th sample, respectively. and are the means of the i-th and j-th features, respectively. Similarly, the discharge covariance matrix elements are calculated based on the multiple discharge linear features in the discharge linear feature set, and the discharge covariance matrix is ​​constructed. The correlation between the multiple charging linear features is calculated based on the charging covariance matrix to obtain the charging feature correlation. The Pearson correlation coefficient can be used for correlation calculation. Assuming that the two features X i and X j The correlation ρ ij The calculation formula is:

[0078]

[0079] Among them, C ij For feature X i and X j The covariance, C ii and C jj They are feature X i and X j The variance of . Similarly, the correlation between multiple discharge linear features is calculated based on the discharge covariance matrix to obtain the discharge feature correlation. According to the charging feature correlation, multiple charging eigenvalues ​​corresponding to multiple charging linear features are calculated. Assuming that the charging feature correlation matrix is ​​R, the eigenvalue λ can be obtained by eigenvalue decomposition iand the corresponding eigenvector v i According to the discharge feature correlation, multiple discharge feature values ​​corresponding to multiple discharge linear features are calculated. According to multiple charging feature values, corresponding charging feature vectors are generated, and the charging feature vectors are normalized to obtain the charging principal component feature weights. Assume that the feature vector is v i , the normalized feature weight w i The calculation formula is:

[0080]

[0081] Among them, ∥v i ∥ is the feature vector v i The Euclidean norm of . Similarly, a corresponding discharge feature vector is generated according to multiple discharge eigenvalues, and the discharge feature vector is normalized by principal components to obtain the discharge principal component feature weight. The charging process data is multiplied by the charging principal component feature weight to obtain multiple charging products, and the multiple charging products are summed to obtain charging feature data. Similarly, the discharge process data is multiplied by the discharge principal component feature weight to obtain multiple discharge products, and the multiple discharge products are summed to obtain discharge feature data. The charging feature data and the discharge feature data are fused, and the charge and discharge cycle test data is obtained by weighted averaging or other methods.

[0082] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0083] (1) Setting the temperature disturbance data of the target battery pack based on the temperature disturbance strategy;

[0084] (2) Calculating the perturbation gradient of the temperature perturbation data to obtain the temperature perturbation gradient parameter;

[0085] (3) Performing a charge and discharge disturbance test on the target battery pack according to the temperature disturbance gradient parameter to obtain recovery process data;

[0086] (4) Aligning the temperature disturbance data and the recovery process data to obtain the initial disturbance test data;

[0087] (5) Perform threshold filtering on the initial disturbance test data to obtain temperature disturbance test data.

[0088] Specifically, determine the specific parameters of the temperature perturbation strategy. These parameters include the range, rate, and duration of temperature change. Through these parameters, simulate the temperature fluctuations that the battery may encounter in actual operation. For example, set the temperature of the battery pack to rise from 25°C to 45°C within a certain period of time, and then drop back to 25°C. Each stage lasts 30 minutes, and the rate of temperature change is 1°C per minute. After setting the temperature perturbation strategy, generate temperature perturbation data. Temperature perturbation data refers to the temperature change data recorded under the set temperature perturbation conditions. Perform perturbation gradient calculation on the temperature perturbation data to obtain the temperature perturbation gradient parameters and determine the rate and amplitude of temperature change. The calculation formula for the temperature perturbation gradient G(t is:

[0089]

[0090] Where G(t) represents the temperature perturbation gradient, Represents the rate of change of temperature over time. By calculating the temperature change rate in each time period, a set of perturbation gradient parameters are obtained. According to the temperature perturbation gradient parameters, the target battery pack is subjected to charge and discharge perturbation test to observe the performance changes of the battery under different temperature conditions. The charge and discharge perturbation test includes multiple charge and discharge operations on the battery under set temperature change conditions, and records the voltage, current, capacity and temperature data during the charge and discharge process. These data will be used as a data set for the battery recovery process to analyze the behavior of the battery under temperature perturbation conditions. The temperature perturbation data and the recovery process data are aligned to obtain the initial perturbation test data. The purpose of data alignment is to match the temperature perturbation data and the charge and discharge recovery process data in time to ensure that there are corresponding temperature and battery performance data at the same time point. The initial perturbation test data is threshold filtered to obtain temperature perturbation test data. The purpose of threshold filtering is to remove noise and outliers in the initial perturbation test data to ensure the accuracy and reliability of the data.

[0091] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0092] (1) performing curve fitting on the charge-discharge cycle test data and the temperature disturbance test data respectively to obtain a charge-discharge cycle test curve and a temperature disturbance test curve;

[0093] (2) performing curve feature recognition on the charge-discharge cycle test curve and the temperature disturbance test curve respectively to obtain a charge-discharge cycle feature point set and a temperature disturbance feature point set;

[0094] (3) performing feature mapping and vector conversion on the charge-discharge cycle feature point set and the temperature disturbance feature point set respectively to obtain a charge-discharge cycle feature vector and a temperature disturbance feature vector;

[0095] (4) defining a first state space model of a charge-discharge cycle characteristic vector and a second state space model of a temperature disturbance characteristic vector, wherein the first state space model includes a first state equation and a first output equation, and the second state space model includes a second state equation and a second output equation;

[0096] (5) creating a first prediction model based on the first state space model, creating a second prediction model based on the second state space model, and defining optimization objectives, solving optimization problems and constraints of the first prediction model and the second prediction model;

[0097] (6) respectively setting the LQR controllers of the first prediction model and the second prediction model to obtain the MPC-LQG model;

[0098] (7) The current balancing control parameters are obtained by analyzing the charge and discharge cycle characteristic vector and the temperature disturbance characteristic vector through the MPC-LQG model.

[0099] Specifically, the collected charge-discharge cycle test data and temperature disturbance test data are preprocessed. These data usually include variables such as voltage, current, and temperature in a time series. The purpose of curve fitting is to convert discrete data points into continuous mathematical expressions for subsequent feature extraction and analysis. Commonly used fitting methods include polynomial fitting, spline fitting, and exponential fitting. Assume that the charge-discharge cycle test data is {(t i ,y i )}, the temperature disturbance test data is {(t j ,z j )}, where t i and t j is the time point, y i and z j They are the corresponding battery performance data. Polynomial fitting is used to obtain the charge and discharge cycle test curve f(t) and the temperature disturbance test curve g(t):

[0100]

[0101] Among them, a k and b lare the coefficients of the fitting polynomial, and n and m are the orders of the polynomial respectively. The curve features are identified for the charge and discharge cycle test curve and the temperature perturbation test curve respectively, and the charge and discharge cycle feature point set and the temperature perturbation feature point set are obtained. Feature points refer to points on the curve that have special significance, such as extreme points, inflection points, and intersection points. These feature points can be identified by analyzing the first-order and second-order derivatives of the curve. The temperature perturbation feature points can also be determined by analyzing the derivatives of the temperature perturbation test curve. Feature mapping and vector conversion are performed on the charge and discharge cycle feature point set and the temperature perturbation feature point set respectively to obtain the charge and discharge cycle feature vector and the temperature perturbation feature vector. Feature mapping is the conversion of the time coordinates of the feature points into the elements of the feature vector. Define the first state space model of the charge and discharge cycle feature vector and the second state space model of the temperature perturbation feature vector. The first state space model includes a first state equation and a first output equation, and the second state space model includes a second state equation and a second output equation. The state space model is a mathematical model used to describe the behavior of a dynamic system. Assume that the state space model of the charge and discharge cycle feature vector is:

[0102] x c (k+1)=A c x c (k)+B c u c (k)+w c (k)

[0103] y c (k) = C c x c (k)+D c u c (k)+v c (k);

[0104] Among them, x c (k) is the state vector, u c (k) is the input vector, y c (k) is the output vector, A c ,B c ,C c ,D c is the system matrix w c (k) and v c (k) are process noise and measurement noise respectively. The state space model of temperature disturbance eigenvector is:

[0105]

[0106] A first prediction model is created based on the first state space model, and a second prediction model is created based on the second state space model, and the optimization objectives, optimization problems and constraints of the first prediction model and the second prediction model are defined. The prediction model predicts the future state of the system through the state space model and determines the optimal control strategy through the optimization algorithm. Assuming that the optimization objective of the first prediction model is to minimize the weighted sum of the state and control input, it can be expressed as:

[0107]

[0108] Among them, J c To optimize the objective function, Q c and R c is the weight matrix, and N is the prediction time domain length. The optimization objective of the second prediction model can be expressed as:

[0109]

[0110] The LQR controllers of the first prediction model and the second prediction model are set respectively to obtain the MPC-LQG model. The LQR controller determines the optimal control strategy of the system by solving the linear quadratic optimization problem. The MPC-LQG model combines the advantages of model predictive control (MPC) and linear quadratic Gaussian (LQG) control, and realizes the current balancing control analysis of the charge and discharge cycle characteristic vector and the temperature disturbance characteristic vector by predicting and optimizing the system state. The current balancing control analysis of the charge and discharge cycle characteristic vector and the temperature disturbance characteristic vector is performed through the MPC-LQG model to obtain the current balancing control parameters. The current balancing control parameters are used to guide the battery management system to achieve optimal current distribution and control under different temperatures and charge and discharge conditions. For example, suppose a battery pack is subjected to a charge and discharge cycle test and a temperature disturbance test, and the following data is obtained:

[0111] Charge and discharge cycle test data:

[0112] {(0,4.2),(10,4.1),(20,4.0),(30,3.9),(40,3.8)};

[0113] Temperature disturbance test data:

[0114] {(0,25),(10,30),(20,35),(30,40),(40,45)};

[0115] The charge-discharge cycle test curve f(t) = 4.2-0.01t and the temperature disturbance test curve g(t) = 25 + 0.5t are obtained by polynomial fitting. After identifying the characteristic points, feature mapping and vector conversion are performed to obtain the charge-discharge cycle characteristic vector v c and the temperature perturbation eigenvector vd Define the state space model and create a prediction model, set up the LQR controller, combine the MPC-LQG model to perform current balance control analysis and finally obtain the current balance control parameters.

[0116] In a specific embodiment, the execution step performs current balancing control analysis on the charge and discharge cycle characteristic vector and the temperature disturbance characteristic vector through the MPC-LQG model to obtain the current balancing control parameter, which may specifically include the following steps:

[0117] (1) performing current balancing control prediction on the charge-discharge cycle characteristic vector by using a first prediction model to obtain a first initial control sequence;

[0118] (2) predicting the current balance control of the temperature disturbance characteristic vector by a second prediction model to obtain a second initial control sequence;

[0119] (3) performing feedback correction on the first initial control sequence through the LQR controller to obtain a first target control sequence, and performing feedback correction on the second initial control sequence to obtain a second target control sequence;

[0120] (4) Perform weighted fusion and parameter mapping on the first target control sequence and the second target control sequence to obtain current balancing control parameters.

[0121] Specifically, the charge-discharge cycle characteristic vector and the first state space model are used for prediction. Assume that the first state space model is:

[0122]

[0123] Among them, x c (k) is the state vector, u c (k) is the control input vector, y c (k) is the output vector, A c ,B c ,C c ,D c is the system matrix, w c (k) and v c (k) are process noise and measurement noise respectively. Based on this model, the model predictive control (MPC) method is used to predict the current balance control. The MPC method predicts the control sequence u of the next N steps at each moment k by solving an optimization problem. c (k) to minimize a cost function. Assume the cost function is:

[0124]

[0125] Among them, Q c and R care the weight matrices of state and control input respectively, and N is the length of prediction time domain. By solving this optimization problem, the first initial control sequence U is obtained c = {u c (0),u c (1),...,u c (N-1)}. The temperature disturbance characteristic vector is predicted by the second prediction model to perform current balancing control and obtain the second initial control sequence. Assume that the second state space model is:

[0126]

[0127] Using the temperature disturbance feature vector and the model for prediction, the second initial control sequence U is obtained by solving the corresponding optimization problem. d = {u d (0),u d (1),...,u d (N-1)}. The first initial control sequence is feedback-corrected by the LQR controller to obtain the first target control sequence. The LQR controller corrects the initial control sequence by minimizing a linear quadratic regulation (LQR) cost function. Assume that the LQR cost function is:

[0128]

[0129] Among them, Q LQR and R LQR is the weight matrix of the LQR controller. By solving the optimization problem, the corrected first target control sequence is obtained: Similarly, the second initial control sequence is feedback-corrected by the LQR controller to obtain the second target control sequence The first target control sequence and the second target control sequence are weighted and fused and parameter mapped to obtain the current balancing control parameters. Assuming the weighted coefficients are α and β, where α+β=1, the weighted fused control sequence U * It can be expressed as

[0130]

[0131] Among them, U c * and U d * They are the first and second target control sequences, respectively, and the final control sequence is obtained after weighted fusion. Parameter mapping is to convert the control sequence into specific current balancing control parameters, which are used to guide the battery management system to achieve optimal current distribution. For example, suppose a battery pack is subjected to a charge and discharge cycle test and a temperature disturbance test, and the following data is obtained:

[0132] Charge and discharge cycle characteristic vector:

[0133] v c =[4.2,4.1,4.0,3.9,3.8];

[0134] Temperature perturbation eigenvector:

[0135] v d =[25,30,35,40,45];

[0136] The current balancing control prediction is performed through the first prediction model to obtain the first initial control sequence:

[0137] U c =[1.0,0.9,0.8,0.7,0.6];

[0138] The current balancing control prediction is performed through the second prediction model to obtain the second initial control sequence:

[0139] U d =[0.5,0.6,0.7,0.8,0.9];

[0140] The first initial control sequence is feedback-corrected by the LQR controller to obtain the first target control sequence

[0141]

[0142] The second initial control sequence is feedback-corrected by the LQR controller to obtain the second target control sequence:

[0143]

[0144] The first target control sequence and the second target control sequence are weighted and fused. Assuming the weight coefficients α = 0.6 and β = 0.4, the fused control sequence is:

[0145]

[0146] Through parameter mapping, the fused control sequence is converted into specific current balancing control parameters. These parameters are used to guide the battery management system to achieve optimal current distribution and control under different charge and discharge cycles and temperature disturbance conditions.

[0147] The battery performance testing method in the embodiment of the present application is described above. The battery performance testing system in the embodiment of the present application is described below. Figure 2 In the embodiment of the present application, an embodiment of the battery performance testing system includes:

[0148] An acquisition module 201 is used to acquire charging performance data of multiple battery cells in a target battery pack, and to construct a current balancing path network of the multiple battery cells according to the charging performance data;

[0149] The calculation module 202 is used to calculate the network characteristics of the current balancing path network to obtain a network characteristic set, and to perform a performance test strategy analysis on the target battery pack according to the network characteristic set to obtain a charge-discharge cycle strategy and a temperature disturbance strategy;

[0150] A cycle test module 203 is used to perform a charge and discharge cycle test on a target battery pack based on a charge and discharge cycle strategy to obtain charge and discharge cycle test data;

[0151] The disturbance test module 204 is used to perform temperature disturbance and recovery tests on the target battery pack based on the temperature disturbance strategy to obtain temperature disturbance test data;

[0152] The analysis module 205 is used to perform current balancing control analysis on the charge-discharge cycle test data and the temperature disturbance test data through the MPC-LQG model to obtain current balancing control parameters.

[0153] Through the collaborative cooperation of the above-mentioned components, by acquiring the charging performance data of multiple battery cells in the target battery pack and constructing a current balancing path network based on these data, the accurate identification and positioning of the battery cells are achieved. This step not only covers the comprehensive measurement and normalization of key parameters such as the internal resistance, SOC (state of charge) and SOH (health state) of the battery cells, but also effectively solves the problem of uneven performance between battery cells by constructing a current balancing path network. It can ensure that the current distribution of each unit in the entire battery pack is uniform during the charging and discharging process, and improve the overall performance and life of the battery pack. The present invention calculates the network characteristics of the current balancing path network in detail, generates a network feature set, and analyzes the performance test strategy based on these characteristics to obtain the charging and discharging cycle strategy and the temperature perturbation strategy. Through the comprehensive analysis of multi-dimensional features such as node degree, path length, clustering coefficient, same-layer degree and inter-layer degree, the accurate evaluation and prediction of battery pack performance is achieved. This method can not only identify potential performance bottlenecks in the battery pack, but also formulate optimized charging and discharging and temperature perturbation strategies, so as to better manage the use and maintenance of the battery pack. By introducing the MPC-LQG model, the current balancing control analysis is performed on the charge and discharge cycle test data and the temperature disturbance test data to obtain the current balancing control parameters. The MPC-LQG model combines the advantages of model predictive control (MPC) and linear quadratic Gaussian (LQG) control, and can adjust the control strategy in real time under dynamic environments. The current distribution and charging process are further optimized through curve fitting, feature recognition and state space model construction. This not only improves the stability and reliability of the battery pack under different working conditions, but also effectively extends the battery life, thereby improving the performance of the battery.

[0154] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0155] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0156] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A battery performance testing method, characterized in that: The battery performance testing method comprises: Acquire charging performance data of a plurality of battery cells in a target battery pack, and construct a current balancing path network of the plurality of battery cells according to the charging performance data; Performing network feature calculation on the current balancing path network to obtain a network feature set, and performing performance test strategy analysis on the target battery pack according to the network feature set to obtain a charge-discharge cycle strategy and a temperature disturbance strategy; Performing a charge-discharge cycle test on the target battery pack based on the charge-discharge cycle strategy to obtain charge-discharge cycle test data; Based on the temperature disturbance strategy, performing temperature disturbance and recovery tests on the target battery pack to obtain temperature disturbance test data; The charge and discharge cycle test data and the temperature disturbance test data are subjected to current balancing control analysis through the MPC-LQG model to obtain current balancing control parameters; specifically comprising: performing curve fitting on the charge and discharge cycle test data and the temperature disturbance test data respectively to obtain a charge and discharge cycle test curve and a temperature disturbance test curve; performing curve feature identification on the charge and discharge cycle test curve and the temperature disturbance test curve respectively to obtain a charge and discharge cycle feature point set and a temperature disturbance feature point set; performing feature mapping and vector conversion on the charge and discharge cycle feature point set and the temperature disturbance feature point set respectively to obtain a charge and discharge cycle feature vector and a temperature disturbance feature vector; defining a first state space model of the charge and discharge cycle feature vector and a second state space model of the temperature disturbance feature vector, wherein the first state space model comprises a first state equation and a first output equation, and the second state space model comprises a second state equation and a second output equation; A first prediction model is created based on the first state-space model, and a second prediction model is created based on the second state-space model, and optimization objectives, optimization problems and constraints of the first prediction model and the second prediction model are defined; LQR controllers of the first prediction model and the second prediction model are set respectively to obtain an MPC-LQG model; a current balancing control prediction is performed on the charge and discharge cycle characteristic vector through the first prediction model to obtain a first initial control sequence; a current balancing control prediction is performed on the temperature disturbance characteristic vector through the second prediction model to obtain a second initial control sequence; feedback correction is performed on the first initial control sequence through the LQR controller to obtain a first target control sequence, and feedback correction is performed on the second initial control sequence to obtain a second target control sequence; weighted fusion and parameter mapping are performed on the first target control sequence and the second target control sequence to obtain current balancing control parameters.

2. The battery performance testing method according to claim 1, characterized in that: The acquiring charging performance data of a plurality of battery cells in a target battery pack and constructing a current balancing path network of the plurality of battery cells according to the charging performance data includes: Identify and locate the battery cells of the target battery pack to obtain multiple battery cells and spatial location information of each battery cell; Collecting historical data of the multiple battery cells to obtain internal resistance data, SOC data and SOH data of each battery cell; The internal resistance data, SOC data and SOH data of each battery cell are normalized to obtain the charging performance data of each battery cell; Initializing nodes of the plurality of battery cells according to the spatial position information to obtain an initial node of each battery cell; Creating an initial current balancing path for each initial node according to the charging performance data; Calculating path weights and assigning nodes to the initial current balancing paths respectively to obtain multiple target current balancing paths and multiple target nodes; A current balancing path network of the plurality of battery cells is constructed according to the plurality of target current balancing paths and the plurality of target nodes.

3. The battery performance testing method according to claim 1, characterized in that: The network characteristic calculation is performed on the current balancing path network to obtain a network characteristic set, and the performance test strategy analysis is performed on the target battery pack according to the network characteristic set to obtain a charge-discharge cycle strategy and a temperature disturbance strategy, including: Calculating the node degree of the current balancing path network to obtain the node degree, and calculating the path length of the current balancing path network to obtain the path length; Calculating a clustering coefficient of the current balancing path network to obtain a clustering coefficient, calculating a same-layer degree of the current balancing path network to obtain a same-layer degree, and calculating an inter-layer degree of the current balancing path network to obtain an inter-layer degree; Generate a corresponding network feature set according to the node degree, the path length, the clustering coefficient, the same-layer degree and the inter-layer degree; Performing correlation calculation on the network feature set and a plurality of preset candidate cycle strategies to obtain a first correlation of each candidate cycle strategy, and selecting a charge and discharge cycle strategy of the target battery pack according to the first correlation; A correlation calculation is performed on the network feature set and a plurality of preset candidate disturbance strategies to obtain a second correlation of each candidate disturbance strategy, and a temperature disturbance strategy of the target battery pack is selected according to the second correlation.

4. The battery performance testing method according to claim 1, characterized in that: The performing a charge-discharge cycle test on the target battery pack based on the charge-discharge cycle strategy to obtain charge-discharge cycle test data includes: Based on the charge-discharge cycle strategy, the target battery pack is subjected to multiple charge tests to obtain charge process data, and the target battery pack is subjected to multiple discharge tests to obtain discharge process data; Performing a linear transformation on the charging process data to obtain a charging linear feature set, and performing a linear transformation on the discharging process data to obtain a discharging linear feature set; Calculating charging covariance matrix elements according to multiple charging linear features in the charging linear feature set, and constructing a charging covariance matrix according to the charging covariance matrix elements; calculating discharging covariance matrix elements according to multiple discharging linear features in the discharging linear feature set, and constructing a discharging covariance matrix according to the discharging covariance matrix elements; Calculating the correlation between the plurality of charging linear features according to the charging covariance matrix to obtain charging feature correlation; calculating the correlation between the plurality of discharging linear features according to the discharging covariance matrix to obtain discharging feature correlation; According to the charging characteristic correlation, a plurality of charging characteristic values ​​corresponding to the plurality of charging linear characteristics are calculated; according to the discharging characteristic correlation, a plurality of discharging characteristic values ​​corresponding to the plurality of discharging linear characteristics are calculated; Generate a corresponding charging feature vector according to the multiple charging feature values, and perform principal component normalization on the charging feature vector to obtain a charging principal component feature weight; generate a corresponding discharging feature vector according to the multiple discharging feature values, and perform principal component normalization on the discharging feature vector to obtain a discharging principal component feature weight; Multiplying the charging process data by the charging main component characteristic weight to obtain a plurality of charging products, and summing the plurality of charging products to obtain charging characteristic data; multiplying the discharging process data by the discharging main component characteristic weight to obtain a plurality of discharging products, and summing the plurality of discharging products to obtain discharging characteristic data; The charging characteristic data and the discharging characteristic data are fused to obtain charging and discharging cycle test data.

5. The battery performance testing method according to claim 1, characterized in that: The step of performing temperature disturbance and recovery tests on the target battery pack based on the temperature disturbance strategy to obtain temperature disturbance test data includes: Setting temperature disturbance data of the target battery pack based on the temperature disturbance strategy; Performing disturbance gradient calculation on the temperature disturbance data to obtain a temperature disturbance gradient parameter; According to the temperature disturbance gradient parameter, a charge and discharge disturbance test is performed on the target battery pack to obtain recovery process data; Performing data alignment on the temperature disturbance data and the recovery process data to obtain initial disturbance test data; Threshold filtering is performed on the initial disturbance test data to obtain temperature disturbance test data.

6. A battery performance testing system, characterized in that: The battery performance testing system comprises: An acquisition module, used to acquire charging performance data of a plurality of battery cells in a target battery pack, and to construct a current balancing path network of the plurality of battery cells according to the charging performance data; A calculation module, used to perform network feature calculation on the current balancing path network to obtain a network feature set, and perform performance test strategy analysis on the target battery pack according to the network feature set to obtain a charge-discharge cycle strategy and a temperature disturbance strategy; A cycle test module, used to perform a charge and discharge cycle test on the target battery pack based on the charge and discharge cycle strategy to obtain charge and discharge cycle test data; A disturbance test module, used to perform temperature disturbance and recovery tests on the target battery pack based on the temperature disturbance strategy to obtain temperature disturbance test data; The analysis module is used to perform current balancing control analysis on the charge and discharge cycle test data and the temperature disturbance test data through the MPC-LQG model to obtain current balancing control parameters; specifically comprising: performing curve fitting on the charge and discharge cycle test data and the temperature disturbance test data respectively to obtain a charge and discharge cycle test curve and a temperature disturbance test curve; performing curve feature recognition on the charge and discharge cycle test curve and the temperature disturbance test curve respectively to obtain a charge and discharge cycle feature point set and a temperature disturbance feature point set; performing feature mapping and vector conversion on the charge and discharge cycle feature point set and the temperature disturbance feature point set respectively to obtain a charge and discharge cycle feature vector and a temperature disturbance feature vector; defining a first state space model of the charge and discharge cycle feature vector and a second state space model of the temperature disturbance feature vector, wherein the first state space model includes a first state equation and a first output equation, and the second state space model includes a second state equation and a second output equation. The invention relates to a method for generating an MPC-LQG model and a method for generating an MPC-LQG model. The method comprises the steps of: generating an MPC-LQG model and a first prediction model based on the first state-space model; creating a first prediction model based on the first state-space model, and creating a second prediction model based on the second state-space model, and defining the optimization objectives of the first prediction model and the second prediction model, solving the optimization problem and the constraint conditions; setting the LQR controllers of the first prediction model and the second prediction model respectively to obtain the MPC-LQG model; performing current balancing control prediction on the charge and discharge cycle characteristic vector through the first prediction model to obtain the first initial control sequence; performing current balancing control prediction on the temperature disturbance characteristic vector through the second prediction model to obtain the second initial control sequence; performing feedback correction on the first initial control sequence through the LQR controller to obtain the first target control sequence, and performing feedback correction on the second initial control sequence to obtain the second target control sequence; performing weighted fusion and parameter mapping on the first target control sequence and the second target control sequence to obtain the current balancing control parameter.

Images