Performance optimization test method and system for new energy batteries

Abstract

The present invention relates to the technical field of battery performance optimization, and discloses a performance optimization test method and system for new energy batteries. The method comprises: collecting multi-channel operating status data of a new energy battery, performing wavelet transform filtering and normalization processing to obtain a standardized feature data matrix; performing operating mode analysis, and obtaining a charge-discharge mode switching strategy through working condition determination calculation; inputting a multi-branch feature extraction network for feature extraction to obtain a parameter-related feature vector; performing grey correlation analysis to obtain a battery state evaluation index; inputting a charge-discharge time optimization model, and obtaining a charge-discharge time prediction curve based on evidence reasoning rules and cohesion factor calculation; and controlling the charge-discharge process in real time according to the charge-discharge time prediction curve and the charge-discharge mode switching strategy. The present invention realizes real-time and precise control of the charge-discharge process, and ensures the safety and stability of the charge-discharge process.

Description

Technical Field

[0001] The present invention relates to the technical field of battery performance optimization, and in particular to a performance optimization testing method and system for new energy batteries. Background Art

[0002] As a key component in the development of clean energy, new energy batteries have long been a focus of industry research on performance optimization and service life extension. Currently, traditional battery performance testing methods rely primarily on fixed charge and discharge strategies and simple threshold control, making it difficult to accurately optimize battery characteristics under different operating conditions. This results in inadequate battery performance and increases the safety risks of overcharging and over-discharging.

[0003] With the development of artificial intelligence technology, deep learning-based battery performance optimization methods are gradually emerging. However, existing methods suffer from problems such as insufficient feature extraction and low pattern recognition accuracy when processing multi-characteristic battery data. In particular, when dealing with the three key data types of charge and discharge characteristics, temperature characteristics, and life characteristics, the lack of an effective feature fusion mechanism makes it difficult to achieve comprehensive assessment and precise control of battery status. Current battery management systems often use a single control strategy for charge and discharge control and are unable to dynamically adjust according to the real-time status of the battery. This leads to unstable battery performance under different temperature environments and operating conditions. At the same time, the lack of accurate prediction and optimization of charge and discharge times significantly affects the battery's efficiency and lifespan. Summary of the Invention

[0004] The present invention provides a performance optimization test method and system for a new energy battery, which realizes real-time and precise control of the charging and discharging process, thereby ensuring the safety and stability of the charging and discharging process.

[0005] In a first aspect, the present invention provides a performance optimization test method for a new energy battery, the performance optimization test method for a new energy battery comprising:

[0006] Collect multi-channel operating status data of new energy batteries, perform wavelet transform filtering and standardization processing, and obtain a standardized feature data matrix;

[0007] Performing an operating mode analysis on the standardized characteristic data matrix, and obtaining a charge-discharge mode switching strategy through operating condition determination and calculation;

[0008] Inputting the standardized feature data matrix into a multi-branch feature extraction network to perform feature extraction to obtain a parameter-associated feature vector;

[0009] Performing grey correlation analysis on the parameter correlation feature vector to obtain a battery state evaluation index;

[0010] Inputting the battery state evaluation index into the charge and discharge time optimization model, and calculating based on the evidence reasoning rule and the cohesion factor to obtain the charge and discharge time prediction curve;

[0011] The charging and discharging process of the new energy battery is controlled in real time according to the charging and discharging time prediction curve and the charging and discharging mode switching strategy.

[0012] In a second aspect, the present invention provides a performance optimization test system for a new energy battery, the performance optimization test system for a new energy battery comprising:

[0013] The acquisition module is used to collect multi-channel operating status data of new energy batteries, and perform wavelet transform filtering and standardization processing to obtain a standardized feature data matrix;

[0014] A mode analysis module is used to perform an operation mode analysis on the standardized characteristic data matrix and obtain a charge-discharge mode switching strategy through operating condition determination and calculation;

[0015] An extraction module, configured to input the standardized feature data matrix into a multi-branch feature extraction network for feature extraction to obtain a parameter-associated feature vector;

[0016] A correlation analysis module is used to perform grey correlation analysis on the parameter correlation feature vector to obtain a battery state evaluation index;

[0017] A calculation module, configured to input the battery state evaluation index into a charge and discharge time optimization model, and calculate based on evidence reasoning rules and cohesion factors to obtain a charge and discharge time prediction curve;

[0018] A control module is used to control the charging and discharging process of the new energy battery in real time according to the charging and discharging time prediction curve and the charging and discharging mode switching strategy.

[0019] The third aspect of the present invention provides a performance optimization testing device for a new energy battery, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the performance optimization testing device for the new energy battery executes the above-mentioned performance optimization testing method for the new energy battery.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the above-mentioned new energy battery performance optimization test method.

[0021] In the technical solution provided by the present invention, the quality of the original data is effectively improved and the noise interference is reduced through multi-channel data acquisition and wavelet transform filtering processing. The multi-branch feature extraction network architecture is adopted to realize the deep feature extraction of three types of data: charge and discharge characteristics, temperature characteristics and life characteristics, thereby improving the feature expression ability and enhancing the adaptability of the model to different working conditions. Based on the grey correlation analysis and fuzzy comprehensive evaluation method, a complete battery status assessment system is constructed, which improves the accuracy and reliability of the status assessment. Through the evidence reasoning rules and cohesion factor calculation, a charge and discharge time optimization model is established, which realizes the accurate prediction of the charge and discharge time and improves the efficiency of the charge and discharge process. Combined with the dynamic boundary constraint conditions and the PID control algorithm, the real-time and precise control of the charge and discharge process is realized, ensuring the safety and stability of the charge and discharge process. The multi-mode switching strategy is adopted to adaptively adjust the control parameters according to different working conditions, thereby improving the flexibility and intelligence level of the battery management system. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 A flow chart of a performance optimization test method for new energy batteries provided in an embodiment of the present application;

[0024] Figure 2 A schematic block diagram of the structure of a performance optimization test system for new energy batteries provided in an embodiment of the present application;

[0025] Figure 3 A schematic block diagram of the structure of the performance optimization test equipment for new energy batteries provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0027] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, combined, or partially merged, so the actual execution order may change based on actual circumstances.

[0028] It should also be understood that the terms used in this specification are for the purpose of describing specific embodiments only and are not intended to limit the present application. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0029] It should be further understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0030] The following describes some embodiments of the present application in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0031] See also Figure 1 , Figure 1 A flow chart of a performance optimization test method for a new energy battery provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the performance optimization test method for new energy batteries provided in the embodiment of the present application includes steps S100 to S600.

[0032] Step S100: collecting multi-channel operating status data of the new energy battery, and performing wavelet transform filtering and normalization processing to obtain a standardized feature data matrix;

[0033] It is understandable that the execution subject of the present invention can be a new energy battery performance optimization test system, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking the server as the execution subject as an example.

[0034] Specifically, for the collection of multi-channel operating status data from new energy batteries, sampling frequencies for different parameters are determined. The sampling frequency for charging voltage, charging current, discharging voltage, and discharging current is set to 100Hz. This high frequency captures rapid changes in the battery during the charging and discharging process. Furthermore, the sampling frequency for surface temperature is set to 10Hz to monitor the battery's thermal characteristics, while the sampling frequency for ambient temperature is set to 1Hz to ensure the impact of the environment on battery performance. Because internal resistance changes relatively slowly, the sampling frequency is set to 0.1Hz to reduce data redundancy and ensure acquisition accuracy. Multi-channel operating status data is transmitted to form a preprocessed dataset. The collected data is integrated and stored via the data transmission channel for subsequent processing. Wavelet transform is applied to the preprocessed dataset to reduce noise. Wavelet transform effectively removes high-frequency noise from the signal, enabling subsequent analysis to be based on clearer data. Wavelet transform can analyze signal characteristics at different scales, improving data reliability. A 3σ outlier detection is performed on the denoised dataset. By calculating the mean and standard deviation of the data, outliers are identified and removed from the dataset to obtain a valid dataset. The valid data set is standardized so that data with different characteristics can be compared at the same scale to obtain a normalized data set. The normalized data set is synchronized with the time series to ensure that the data of each parameter can correspond to each other in the time dimension, forming a synchronized data set. Through time synchronization, the time misalignment problem caused by different sampling frequencies can be eliminated, so that subsequent analysis can be based on accurate and consistent data. After synchronization, the data is classified and organized according to the charge and discharge characteristics, temperature characteristics, and life characteristics to form a feature data set. The feature data set is subjected to matrix reconstruction operation, and the charge and discharge characteristic data, temperature characteristic data, and life characteristic data are rearranged according to the time sequence and parameter correspondence to form a standardized feature data matrix.

[0035] Step S200: Analyze the operating mode of the standardized characteristic data matrix and obtain a charge-discharge mode switching strategy through operating condition determination and calculation;

[0036] Specifically, the standardized feature data matrix is used to extract charge and discharge parameter features. Key battery parameters during the charge and discharge processes, such as charge voltage, charge current, discharge voltage, and discharge current, are identified and extracted to form a standard charge and discharge mode parameter set. Simultaneously, temperature data is subjected to threshold analysis to identify the battery's operating characteristics under different temperature conditions and extract a temperature regulation mode parameter set to help understand the battery's behavior in high and low temperature environments. The cycle count and capacity data in the standardized feature data matrix are analyzed to obtain a lifespan optimization mode parameter set. By integrating the standard charge and discharge mode parameter set, the temperature regulation mode parameter set, and the lifespan optimization mode parameter set, a mode feature vector space is constructed, resulting in a multi-mode feature matrix that fully reflects the battery's operating characteristics under different operating conditions. Based on this, operating condition classification calculations are performed, and the Euclidean distance between the current battery state parameters and the mode feature vectors is compared to obtain operating condition matching data. Euclidean distance is an important method for assessing the similarity between different data points. By calculating the distance between the current state and each mode, the operating mode that most closely matches the current operating condition is identified. Based on the matching data, a mode analysis is performed, and the operating mode with the highest matching degree is selected as the optimal mode for the current operating conditions. This ensures that the battery operates optimally in different usage scenarios, improving its performance and safety. Smooth transition processing is performed based on the mode switching sequence to ensure the continuity of current and voltage parameters during mode switching. Constrained calculations are performed on the current and voltage parameters to ensure that no sudden changes occur during mode switching, protecting the battery's internal structure and the stability of the battery management system. These calculations generate a transition control sequence, ensuring that the battery can smoothly transition to the new operating state when switching between charge and discharge modes. The transition control sequence is reconstructed according to the timing relationship to form a charge and discharge mode switching strategy. This strategy includes the mode switching criteria and a strategy execution table for the control parameters.

[0037] Step S300: input the standardized feature data matrix into a multi-branch feature extraction network to extract features and obtain a parameter-related feature vector;

[0038] Specifically, the standardized feature data matrix is separated, with the charge-discharge characteristic data, temperature characteristic data, and life characteristic data being fed into separate feature branch networks. This separation ensures that the features of different characteristics can be processed and analyzed independently, extracting more targeted feature information. The charge-discharge characteristic data from the first feature branch network is processed using a sliding window, with a window length of 1000 data points and a sliding step of 100 data points. This processing method enables segmented analysis of the time series data, capturing the dynamic characteristics of the charge and discharge process. The charge-discharge characteristic matrix is obtained by calculating the mean, standard deviation, peak, and valley values of the data within each window. Fast Fourier transform (FFT) is used to process the temperature characteristic data to analyze temperature variations in the frequency domain. By selecting the amplitude spectrum within the 0-50 Hz frequency band, the main components of temperature fluctuations are captured. Spectral peak detection and energy distribution calculations are used to obtain the temperature characteristic matrix. For the life characteristic data, wavelet transform decomposition is used, with wavelet coefficients at five scales selected. This technique effectively captures local features and changing trends in the data. By calculating statistical moments and energy entropy, a lifetime feature matrix is derived, providing an important quantitative indicator of the battery's service life. In the feature extraction layer construction, the charge-discharge feature matrix is input into the first feature extraction layer, which consists of three convolutional layers. Each convolutional layer uses the ReLU activation function and batch normalization, with kernel sizes set to 3×3, 5×5, and 7×7, respectively. This multi-level convolution operation effectively extracts deep information from the charge-discharge characteristics, forming a charge-discharge deep feature. The temperature feature matrix is input into the second feature extraction layer, which consists of three one-dimensional convolutional layers, with a max pooling layer and a LeakyReLU activation function, and a pooling size of 2. This layer is designed to extract the temporal characteristics of the temperature characteristic data through a combination of convolution and pooling, resulting in a temperature deep feature. The lifetime feature matrix is input into the third feature extraction layer, which consists of two fully connected layers and a transformer encoder layer with four attention heads, which effectively captures long-range dependencies and complex feature interactions, generating a lifetime deep feature. Three-way self-attention is calculated for the charge-discharge depth features, temperature depth features, and lifespan depth features, and the multi-head attention mechanism is used to build correlations between the features. This mechanism identifies the importance of each feature in a specific situation and calculates a three-way attention weight matrix. Using this weight matrix, a weighted fusion of the charge-discharge depth features, temperature depth features, and lifespan depth features is performed. Residual connections and layer normalization allow the different features to complement each other during the fusion process, forming a fused feature tensor. To extract information, the fused feature tensor is subjected to dimensionality reduction using a fully connected layer to obtain a parameter-correlated feature vector.

[0039] Step S400: performing grey correlation analysis on the parameter correlation feature vector to obtain a battery status evaluation index;

[0040] Specifically, the gray correlation coefficient is calculated for the parameter-association eigenvector, quantifying the relationship between different parameters by applying a specific formula. The parameter correlation coefficient is calculated using the formula γ=(1+|ξ|) / (1+|ξ|+|Δi-Δj|), where γ represents the parameter correlation coefficient, ξ is the resolution coefficient, which is set to 0.5, and Δi and Δj represent the rates of change of the i-th parameter and the j-th parameter, respectively. This calculation process generates a correlation matrix, reflecting the degree of mutual influence between the parameters. Based on the correlation matrix, a temperature-depth state function is constructed, expressed as S=α·exp(-β·T)+λ·DOD², where S is the temperature-depth state function, T is the temperature, DOD represents the depth of discharge, and α, β, and λ are the coefficients of the state function. This state function captures the relationship between temperature and depth of discharge. The correlation matrix is subjected to eigenvalue decomposition, and principal component analysis is used to extract the main influencing factors. By identifying the main components in the data, redundant information can be effectively reduced and the parameters that have the greatest impact on the battery state can be extracted. Based on the extracted principal components, the weights of each parameter are calculated and dynamic weight coefficients are obtained. On this basis, a fuzzy evaluation matrix is constructed by combining the state function coefficients and dynamic weight coefficients, with charging efficiency, discharge capacity, temperature uniformity, and cycle life as evaluation factors. The fuzzy evaluation matrix quantifies the complex relationships between different evaluation factors, forming a multidimensional evaluation framework. A hierarchical analysis is performed on the fuzzy relationship matrix, and the weight vectors for each evaluation factor are calculated by establishing a judgment matrix. Consistency checks ensure the rationality of the weights, making the evaluation process more scientific and effective. Throughout this process, a reasonable weight allocation ensures that the contribution of each evaluation factor to the overall evaluation is fully reflected. The comprehensive weight vector is combined with the fuzzy relationship matrix, and fuzzy synthesis is performed using the weighted average method to determine the evaluation level. Using the maximum membership principle, the evaluation results are effectively converted into a specific battery state assessment index, providing a quantitative basis for the battery's health.

[0041] Step S500: Input the battery state evaluation index into the charge and discharge time optimization model, and obtain a charge and discharge time prediction curve based on the evidence reasoning rule and the cohesion factor calculation;

[0042] Specifically, the evidence reasoning analysis of the battery state evaluation index is carried out, and the charge and discharge time cohesion factor is constructed by calculating the formula CF=Σ(wi·ri) / Σwi. CF represents the charge and discharge time cohesion factor, wi is the weight coefficient of the charge time and discharge time, and ri represents the reliability of the time parameters during the battery charge and discharge process. Through calculation, the time characteristic correlation coefficient is obtained. According to the time characteristic correlation coefficient, the dynamic boundary constraint condition is calculated. Through the function Tchg min =T0·exp(-μt) to calculate the minimum charging time, through the function Tchg max =T1·exp(-νt) to calculate the maximum charging time, where T0 and T1 are the initial minimum and maximum charging time values, respectively, t represents the number of battery cycles, and μ and ν are the charging time attenuation coefficients. In this way, the charging time constraint function is obtained, which reflects the trend of charging time changing with the number of battery cycles. The discharge time is mapped to the charging time constraint function. Based on the relationship between battery capacity and discharge rate, a discharge time constraint function is constructed, specifically expressed as Tdis min =k1·Tchg min and Tdis max =k2·Tchg max , where Tdis min is the minimum discharge time, Tdis maxis the maximum discharge time, while k1 and k2 are the minimum and maximum discharge time coefficients. This approach forms a complete time constraint. The complete time constraint is input into the time optimizer. Within the time optimizer, the number of discrete points for the charge and discharge times is set, typically 100. These discrete points represent time sampling points during the charge and discharge processes. A particle swarm optimization algorithm is used to solve the problem and obtain an initial time series. This initial time series is dynamically optimized to construct a time optimization objective function: F = w1·Tchg+w2·Tdis+w3·ΔT, where F is the objective function value, Tchg is the charge time, Tdis is the discharge time, and ΔT represents the temperature change during the charge and discharge processes. w1, w2, and w3 are the weights for the charge time, discharge time, and temperature change, respectively. This ensures that the impact of various factors on time prediction is fully considered during the optimization process, forming a comprehensive optimization objective. Based on this objective function, an iterative calculation is performed, with a set number of particles of 50 and a maximum number of iterations of 500, where each particle represents a candidate solution. The inertia weight is set to 0.8, and the learning factor is set to 2.0. The former indicates the degree to which the particle maintains its current velocity, while the latter reflects the degree to which the particle learns toward the optimal solution. This process gradually approaches the optimal time solution through multiple iterations, ensuring that the final time prediction result is highly accurate and reliable. After obtaining the optimal time solution, the sequence is reconstructed and a continuous time curve is generated using cubic spline interpolation. This method smoothly connects the various time sampling points, forming a more continuous and natural charge and discharge time prediction curve. The continuous time curve is dynamically corrected, calibrated and smoothed based on real-time changes in temperature and state of charge. This correction process aims to ensure the accuracy of the charge and discharge time prediction curve, so that it can reflect the battery's actual operating state and obtain a charge and discharge time prediction curve.

[0043] Step S600: Control the charge and discharge process of the new energy battery in real time according to the charge and discharge time prediction curve and the charge and discharge mode switching strategy.

[0044] Specifically, the charge and discharge time prediction curve is subjected to time-segmented control. By analyzing the charging and discharging time ranges, corresponding control intervals are set to generate a segmented control sequence. The charge and discharge mode switching strategy is analyzed, and the mode switching criteria and control parameters corresponding to the current operating conditions are extracted from the strategy execution table. The extracted mode control instructions provide the specific actions required for the current battery state. Parameter matching is performed by combining the segmented control sequence and the mode control instructions. The charging current, charge cutoff voltage, and discharge cutoff voltage are configured online to ensure dynamic adjustment based on the current operating conditions, generating real-time control parameters. This online configuration enables the system to quickly respond to changes in battery status and achieve more precise control. Safety constraints are imposed on the real-time control parameters. Batteries face various risks during the charging and discharging process, such as overcharging, overdischarging, and overheating. By setting safety control parameters, the system ensures that charging and discharging operations remain within a safe range, protecting the overall performance and service life of the battery. A feedback controller is constructed based on the safety control parameters. A PID control algorithm is used for closed-loop control of the charging current. The proportional coefficient Kp is set to 0.8, the integral coefficient Ki is set to 0.1, and the differential coefficient Kd is set to 0.05. These parameters work together in the control algorithm to dynamically adjust the charging current based on real-time feedback, ensuring stability and efficiency during the charging and discharging process. The PID control algorithm continuously compares the setpoint with the actual value, automatically adjusting the output signal to achieve optimal control. The control output signal undergoes power conversion to achieve specific control of the battery charging and discharging process. This conversion process uses PWM (pulse width modulation) to control the on-time of the power transistors, generating charge and discharge execution instructions. PWM modulation is a highly efficient power control technology that precisely controls current by adjusting the on-time, enabling real-time regulation of the battery charging and discharging process. This method allows the system to efficiently manage battery charging and discharging, ensuring both charging efficiency and battery safety.

[0045] In the embodiment of the present invention, through multi-channel data acquisition and wavelet transform filtering processing, the quality of the original data is effectively improved and noise interference is reduced. The multi-branch feature extraction network architecture is adopted to realize deep feature extraction of three types of data: charge and discharge characteristics, temperature characteristics and life characteristics, thereby improving the feature expression ability and enhancing the adaptability of the model to different working conditions. Based on grey correlation analysis and fuzzy comprehensive evaluation method, a complete battery status assessment system is constructed, which improves the accuracy and reliability of status assessment. Through evidence reasoning rules and cohesion factor calculation, a charge and discharge time optimization model is established, which realizes accurate prediction of charge and discharge time and improves the efficiency of the charge and discharge process. Combined with dynamic boundary constraints and PID control algorithm, real-time and precise control of the charge and discharge process is realized, ensuring the safety and stability of the charge and discharge process. A multi-mode switching strategy is adopted to adaptively adjust control parameters according to different working conditions, thereby improving the flexibility and intelligence level of the battery management system.

[0046] In a specific embodiment, the process of executing step S100 may specifically include the following steps:

[0047] The charging voltage, charging current, discharging voltage, and discharging current of new energy batteries are sampled at a frequency of 100 Hz, the surface temperature at a frequency of 10 Hz, the ambient temperature at a frequency of 1 Hz, and the internal resistance at a frequency of 0.1 Hz to obtain multi-channel operating status data;

[0048] Transmitting multi-channel operating status data to obtain a preprocessed data set, and performing wavelet transform processing on the preprocessed data set to obtain a noise reduction data set;

[0049] Perform 3σ outlier detection on the denoised dataset to obtain a valid dataset, and then perform standardization on the valid dataset to obtain a normalized dataset;

[0050] Performing time series synchronization processing on the normalized data set to obtain a synchronized data set, and classifying and arranging the synchronized data set according to charge and discharge characteristics, temperature characteristics, and life characteristics to obtain a feature data set;

[0051] A matrix reconstruction operation is performed on the characteristic data set, and the charge and discharge characteristic data, temperature characteristic data and life characteristic data are rearranged according to the time sequence and parameter correspondence to obtain a standardized characteristic data matrix.

[0052] Specifically, determine the sampling frequency. The charging voltage, charging current, discharge voltage and discharge current are collected at a sampling frequency of 100Hz, and 100 data points are obtained per second. The choice of this frequency is based on the characteristics of rapid changes in current and voltage during battery charging and discharging, which can effectively capture its instantaneous state. At the same time, the surface temperature will be collected at a frequency of 10Hz to reduce the amount of data without losing important information. The ambient temperature is collected at a frequency of 1Hz to reflect the slower changes in the impact of the environment on battery performance. The sampling frequency of the internal resistance value is set to 0.1Hz. The internal resistance changes relatively slowly and does not require too frequent data points. This series of sampling forms a multi-channel operating status data set, including the key performance indicators of the battery. The collected multi-channel operating status data is data-transmitted from the sensor or data acquisition device to the computer system to form a pre-processed data set. The pre-processed data set is processed by wavelet transform to obtain a noise reduction data set. Wavelet transform is an effective signal processing technology that helps separate noise from the signal. The formula is expressed as:

[0053] ;

[0054] in, is the transformed signal, is the wavelet function, is the original signal. This processing effectively removes high-frequency noise while retaining the important low-frequency signal. After completing the wavelet transform, the denoised dataset is obtained. A 3σ outlier test is performed on the denoised dataset to filter out the valid dataset. The 3σ principle is based on statistics and states that in a normal distribution, approximately 99.73% of the data should fall between ±3 standard deviations of the mean. The test process is expressed as follows:

[0055] ;

[0056] in, is the mean of the data set, is the standard deviation. Through this step, a valid data set is obtained. The valid data set is standardized to eliminate the influence between different dimensions and scales to obtain a normalized data set. The standardization process uses the following formula:

[0057] ;

[0058] in, is the normalized value, is the original data, is the data mean, is the standard deviation. After standardization, the normalized dataset is synchronized to ensure the consistency of all data on the time axis, forming a synchronized dataset. If the timestamps of different data are inconsistent, interpolation is performed to align all data to the same time base. The formula for time synchronization is expressed as:

[0059] ;

[0060] in, represents the interpolation function, For different time points. After the synchronization data set is completed, the synchronization data set is classified and sorted according to the charge and discharge characteristics, temperature characteristics and life characteristics to obtain a feature data set. The classification process can ensure the pertinence and effectiveness of subsequent analysis, so that each feature set can reflect the performance of the battery under different working conditions. The matrix reconstruction operation is performed on the feature data set, and the charge and discharge characteristic data, temperature characteristic data and life characteristic data are rearranged according to the time sequence and parameter correspondence to obtain a standardized feature data matrix. The process of reconstructing the matrix is expressed by the following formula:

[0061] ;

[0062] in, represents the reconstructed feature data matrix, Indicates the charge and discharge characteristics, Indicates temperature characteristics, Represents the life characteristics. All these data are integrated in chronological order so that each row corresponds to a different characteristic value at the same time point.

[0063] In a specific embodiment, the process of executing step S200 may specifically include the following steps:

[0064] Perform charge and discharge parameter feature extraction on the standardized feature data matrix to obtain a standard charge and discharge mode parameter set, and perform threshold analysis on the temperature data in the standardized feature data matrix to obtain a temperature regulation mode parameter set; analyze the number of cycles and capacity data in the standardized feature data matrix to obtain a life optimization mode parameter set;

[0065] According to the standard charge and discharge mode parameter set, the temperature adjustment mode parameter set and the life optimization mode parameter set, a mode feature vector space is constructed to obtain a multi-mode feature matrix;

[0066] Perform operating condition classification calculations on the multi-mode feature matrix, obtain operating condition matching data based on the Euclidean distance between the current battery state parameters and the mode feature vector, and perform mode analysis based on the operating condition matching data. Select the operating mode with the highest matching degree as the optimal mode for the current operating condition, and obtain the mode switching sequence.

[0067] Based on the mode switching sequence, smooth transition processing is performed, and continuity constraint calculation is performed on the current and voltage parameters to obtain the transition control sequence. The transition control sequence is reconstructed according to the timing relationship to generate the charge and discharge mode switching strategy. The charge and discharge mode switching strategy includes a strategy execution table of mode switching criteria and control parameters.

[0068] Specifically, the charge and discharge parameter features of the standardized characteristic data matrix are extracted, and key information is extracted from the matrix to construct a standard charge and discharge mode parameter set. This process involves analyzing multiple indicators such as charge and discharge voltage, charge and discharge current, charge time, and discharge time. The charge and discharge parameter features are obtained by calculating the mean, maximum, minimum, and standard deviation of the charge and discharge voltage and current. The formula is expressed as:

[0069] ;

[0070] in, is the characteristic mean, is the sample size, It is In a similar way, the fluctuation of charge and discharge voltage and current, that is, the standard deviation, is calculated:

[0071] ;

[0072] On this basis, a standard charge and discharge mode parameter set is established, which includes parameters such as charging efficiency, discharging efficiency, charging voltage range, and discharging voltage range. At the same time, threshold analysis is performed on the temperature data in the standardized feature data matrix to identify the temperature adjustment mode parameter set. A temperature threshold is set, and data points exceeding this threshold are considered abnormal or require adjustment. Threshold analysis is expressed using the following formula:

[0073] ;

[0074] in, is the temperature data, is the preset temperature threshold. This method identifies temperature data requiring special attention and constructs a corresponding temperature regulation mode parameter set. Similar analysis is performed on the cycle count and capacity data in the standardized feature data matrix to obtain a life optimization mode parameter set. This includes calculating the average cycle count, maximum capacity, minimum capacity, and the relationship between cycle count and capacity. For example, the relationship between cycle count and capacity is represented by the following linear regression model:

[0075] ;

[0076] in, is the capacity, is the number of cycles, and is the regression coefficient, estimated using the least squares method to construct a model parameter set related to life optimization. Based on the standard charge and discharge model parameter set, the temperature adjustment model parameter set, and the life optimization model parameter set, a model feature vector space is constructed to form a multi-model feature matrix. All model parameter sets are integrated to form a high-dimensional feature space. All feature parameters are combined into a matrix representation, which is as follows:

[0077] ;

[0078] in, 、 、 Represent the parameters of the standard charge and discharge mode, the temperature regulation mode, and the life optimization mode, respectively. The multi-mode feature matrix is used to classify the operating conditions. The Euclidean distance between the current battery state parameters and the mode feature vector is used to obtain the operating condition matching data. The Euclidean distance calculation formula is:

[0079] ;

[0080] in, and are the current battery state parameters and the first elements. By calculating the matching degree of all modes, the operating mode with the highest matching degree is selected as the optimal mode for the current working condition, and the mode switching sequence is obtained. The generation of the mode switching sequence involves the smooth transition processing of the current and voltage parameters during the charging and discharging process. In order to ensure the stability during the mode switching, the continuity constraint calculation is performed. By setting the maximum change rate of current and voltage, it is ensured that no sudden changes occur during the switching process. For example, setting the current change rate not to exceed The condition is expressed as:

[0081] ;

[0082] in, and are the current values after and before switching, respectively. This constraint yields a transition control sequence, which is reconstructed according to the timing relationship to generate a charge-discharge mode switching strategy. The final charge-discharge mode switching strategy includes a strategy execution table for mode switching criteria and control parameters, listing the charge-discharge mode, parameter settings, and switching conditions to be executed under specific operating conditions.

[0083] In a specific embodiment, the process of executing step S300 may specifically include the following steps:

[0084] Performing data separation on the standardized characteristic data matrix, inputting the charge and discharge characteristic data into the first characteristic branch network, inputting the temperature characteristic data into the second characteristic branch network, and inputting the life characteristic data into the third characteristic branch network;

[0085] Perform sliding window processing on the charge and discharge characteristic data of the first feature branch network. The window length is set to 1000 data points and the sliding step is set to 100 data points. The charge and discharge characteristic matrix is obtained by calculating the mean, standard deviation, peak value and valley value.

[0086] Perform fast Fourier transform on the temperature characteristic data of the second characteristic branch network, select the amplitude spectrum in the 0-50 Hz frequency band, and obtain the temperature characteristic matrix through spectrum peak detection and energy distribution calculation;

[0087] The life characteristic data of the third characteristic branch network are decomposed by wavelet transform, and the wavelet coefficients of 5 scales are selected. The life characteristic matrix is obtained by statistical moment and energy entropy calculation.

[0088] The charge-discharge feature matrix is input into the first feature extraction layer. The first feature extraction layer contains three convolutional layers. Each convolutional layer uses the ReLU activation function and the batch normalization layer. The convolution kernel sizes are 3×3, 5×5, and 7×7 respectively to obtain the charge-discharge depth features.

[0089] The temperature feature matrix is input into the second feature extraction layer, which contains three one-dimensional convolutional layers. Each convolutional layer is configured with a maximum pooling layer and a LeakyReLU activation function, and the pooling size is set to 2 to obtain the temperature depth feature.

[0090] The lifespan feature matrix is input into the third feature extraction layer, which contains two fully connected layers and one transformer encoder layer. The transformer encoder layer contains four attention heads to obtain the lifespan deep feature.

[0091] Perform three-way self-attention calculation on the charge and discharge depth features, temperature depth features, and life depth features, build the correlation between features through the multi-head attention mechanism, and obtain the three-way attention weight matrix;

[0092] The charge and discharge depth features, temperature depth features, and life depth features are weightedly fused according to the three-way attention weight matrix. The fused feature tensor is obtained through residual connection and layer normalization. The fused feature tensor is then reduced in dimension through the fully connected layer to obtain the parameter-associated feature vector.

[0093] Specifically, the standardized feature data matrix is separated, and the charge and discharge characteristic data, temperature characteristic data and life characteristic data are respectively input into different feature branch networks. The charge and discharge characteristic data are input into the first feature branch network, the temperature characteristic data are input into the second feature branch network, and the life characteristic data will enter the third feature branch network. Special processing methods are used for different types of data to improve the accuracy and effectiveness of feature extraction. For the first feature branch network, the charge and discharge characteristic data are processed by sliding window. The window length is set to 1000 data points, and the sliding step is set to 100 data points. During processing, a window containing 1000 data points is slid to the next position each time to extract the time series features in the data. In each window, the mean, standard deviation, peak value and valley value are calculated to form a charge and discharge characteristic matrix. For the temperature characteristic data of the second feature branch network, a fast Fourier transform is performed to analyze the performance of the temperature data in the frequency domain. The amplitude spectrum in the 0-50Hz frequency band is selected to obtain the frequency characteristics of the temperature signal. The formula for Fourier transform is:

[0094] ;

[0095] in, is the frequency domain signal, is the time domain signal, is the total number of data points. By detecting spectral peaks and calculating energy distribution, a temperature characteristic matrix is obtained, which captures the periodic characteristics of temperature changes over time. For the life characteristic data of the third characteristic branch network, wavelet transform is used for decomposition, and wavelet coefficients at five scales are selected for analysis. The wavelet transform is described by the following formula:

[0096] ;

[0097] in, are the wavelet coefficients, is the scale parameter, is the translation parameter, is the original signal, is the mother wavelet function. By calculating the statistical moment and energy entropy, the life feature matrix is extracted from the wavelet coefficients to quantify the change in battery life. The charge and discharge feature matrix is input into the first feature extraction layer. This layer contains 3 convolutional layers, each of which uses the ReLU activation function and the batch normalization layer. The convolution kernel sizes of the convolutional layers are 3×3, 5×5, and 7×7, respectively. Through this setting, features of different scales are extracted to obtain charge and discharge depth features. The ReLU activation function is defined as:

[0098] ;

[0099] Effectively introduce nonlinearity while reducing the problem of vanishing gradients. The temperature feature matrix is input to the second feature extraction layer, which contains three one-dimensional convolutional layers. Each convolutional layer is followed by a maximum pooling layer and a LeakyReLU activation function, with the pooling size set to 2. The LeakyReLU activation function is defined as:

[0100] ;

[0101] in, is a small constant. This processing method better captures changes in temperature features. The lifetime feature matrix is input into the third feature extraction layer, which consists of two fully connected layers and one transformer encoder layer. In the transformer encoder layer, four attention heads are used to capture the relationship between features. The transformer's attention mechanism is described by the following formula:

[0102] ;

[0103] in, 、 、 Represents query, key and value respectively, is the dimension of the key. Through the multi-head attention mechanism, a three-way attention weight matrix is obtained, which reflects the correlation between the charge and discharge features, temperature features, and lifespan features. The three-way attention weight matrix is used to perform weighted fusion of the charge and discharge depth features, temperature depth features, and lifespan depth features. Through residual connection and layer normalization, the fused feature tensor is obtained. The calculation of the fused feature tensor is expressed as:

[0104] ;

[0105] in, is the input feature, and LayerNorm is the layer normalization operation. The fused feature tensor is reduced in dimension through the fully connected layer to obtain the parameter-associated feature vector.

[0106] In a specific embodiment, the process of executing step S400 may specifically include the following steps:

[0107] The grey correlation degree is calculated for the parameter association eigenvector, and the correlation coefficient between parameters is calculated based on the formula γ=(1+|ξ|) / (1+|ξ|+|Δi-Δj|), where γ is the correlation coefficient between parameters, ξ is the resolution coefficient, which is 0.5, Δi and Δj are the change rates of the i-th parameter and the j-th parameter, respectively, to obtain the correlation matrix;

[0108] According to the correlation matrix, the temperature-depth state function S=α·exp(-β·T)+λ·DOD² is constructed, where S is the temperature-depth state function, T is the temperature value, DOD is the discharge depth, and α, β, and λ are the state function coefficients;

[0109] Perform eigenvalue decomposition on the correlation matrix, extract the main influencing factors through principal component analysis, calculate the weight of each parameter based on the contribution rate, and obtain the dynamic weight coefficient;

[0110] A fuzzy evaluation matrix is constructed based on the state function coefficient and the dynamic weight coefficient, and the charging efficiency, discharge capacity, temperature uniformity and cycle life are set as evaluation factors to obtain the fuzzy relationship matrix.

[0111] Perform hierarchical analysis on the fuzzy relationship matrix, calculate the weight vector of each evaluation factor by establishing a judgment matrix, use consistency test to ensure the rationality of the weight, and obtain the comprehensive weight vector;

[0112] The comprehensive weight vector and the fuzzy relationship matrix are compounded and the weighted average method is used for fuzzy synthesis. The evaluation level is determined by the maximum membership principle to obtain the fuzzy evaluation result, and the battery state evaluation index is calculated based on the fuzzy evaluation result.

[0113] Specifically, the grey correlation coefficient of the parameter correlation feature vector is calculated. The calculation formula is expressed as:

[0114] ;

[0115] in, It is Parameters and The correlation coefficient of the parameters, is the resolution coefficient, usually set to 0.5. and Representing the Parameters and The rate of change of a parameter. The rate of change is calculated using the following formula:

[0116] ;

[0117] in, is the parameter value at the current moment, is the parameter value at the previous moment. By calculating multiple parameters, a correlation matrix containing the relationship between all parameters is obtained. The temperature-depth state function matrix is constructed based on the correlation matrix. , which has the form:

[0118] ;

[0119] in, is the temperature-depth state function, Represents the current temperature value, is the depth of discharge, 、 and are the coefficients of the state function. These coefficients are obtained by fitting experimental data or optimizing algorithms to ensure that the model can accurately reflect the performance state of the battery. For practical applications, it is assumed that 、 and , then use these parameters and the current temperature and discharge depth to calculate the corresponding state value Perform eigenvalue decomposition on the correlation matrix to extract the main influencing factors. Through principal component analysis, identify the main components in the data and calculate the weight of each parameter based on the contribution rate. Calculate the eigenvalue and eigenvector of the correlation matrix and select the top eigenvectors, calculate the contribution rate of each eigenvector, and the contribution rate is expressed by the following formula:

[0120] ;

[0121] in, It is The contribution rate of the eigenvectors, is its corresponding eigenvalue, is the total number of eigenvalues. Through calculation, the dynamic weight coefficient is obtained to reflect the influence of each parameter on the system state. A fuzzy evaluation matrix is constructed based on the state function coefficient and the dynamic weight coefficient. Charging efficiency, discharge capacity, temperature uniformity and cycle life are set as evaluation factors to form a fuzzy relationship matrix. . This matrix is represented as:

[0122] ;

[0123] in, Indicates the The first judging factor and the To ensure the rationality of the fuzzy relationship matrix, a hierarchical analysis is performed to calculate the weight vector of each evaluation factor by establishing a judgment matrix. The construction of the judgment matrix is based on expert judgment or historical data, and the relative importance of each pair of evaluation factors is scored to construct the judgment matrix. , and calculate the eigenvalues and eigenvectors, and use consistency test to ensure the rationality of the weights. The commonly used consistency index is the consistency ratio , and its calculation formula is:

[0124] ;

[0125] in, is the consistency indicator, is a random consistency indicator, is the order of the judgment matrix. If 0.1, the judgment matrix has good consistency. Perform a composite operation on the comprehensive weight vector and the fuzzy relationship matrix. Fuzzy synthesis is performed using the weighted average method, which is expressed as the following formula:

[0126] ;

[0127] in, It is The comprehensive score of the evaluation factors, It is The weight of the judging factors, It is an element in the fuzzy relationship matrix. Through calculation, the fuzzy evaluation results of each evaluation factor are obtained. The evaluation level is determined by the maximum membership principle, and the evaluation factor with the largest comprehensive score is selected as the final evaluation result. By summarizing all the evaluation results, the battery status evaluation index is calculated. , this index can effectively reflect the overall performance of the battery. The calculation formula of the battery status evaluation index is expressed as:

[0128] ;

[0129] In a specific embodiment, the process of executing step S500 may specifically include the following steps:

[0130] Evidence reasoning analysis is performed on the battery state assessment index. The charge and discharge time convergence factor is constructed by calculating the formula CF=Σ(wi·ri) / Σwi, where CF is the charge and discharge time convergence factor, wi is the weight coefficient of the charge time and discharge time, and ri is the reliability of the time parameter during the battery charge and discharge process. The time characteristic correlation coefficient is obtained.

[0131] The dynamic boundary constraint conditions are calculated based on the time characteristic correlation coefficient, and the function Tchg min =T0·exp(-μt) to calculate the minimum charging time, through the function Tchg max =T1·exp(-νt) to calculate the maximum charging time, where Tchg min is the minimum charging time, Tchg max is the maximum charging time, T0 is the initial minimum charging time value, T1 is the initial maximum charging time value, t is the number of battery cycles, μ and ν are the charging time attenuation coefficients, and the charging time constraint function is obtained;

[0132] The discharge time constraint function is mapped to the charge time constraint function, and the discharge time constraint function Tdis is constructed based on the relationship between battery capacity and discharge rate. min =k1·Tchg min and Tdis max =k2·Tchg max , where Tdis min is the minimum discharge time, Tdis max is the maximum discharge time, k1 is the minimum discharge time coefficient, k2 is the maximum discharge time coefficient, and the complete time constraint condition is obtained;

[0133] The complete time constraint condition is input into the time optimizer, and the number of discrete points of charging time is set to 100, and the number of discrete points of discharging time is set to 100. The discrete points represent the time sampling points in the charging and discharging process. The particle swarm optimization algorithm is used to solve the problem and obtain the initial time series.

[0134] The initial time series is dynamically optimized to construct the time optimization objective function F=w1·Tchg+w2·Tdis+w3·ΔT, where F is the objective function value, Tchg is the charging time, Tdis is the discharging time, ΔT is the temperature change during the charging and discharging process, w1 is the charging time weight, w2 is the discharging time weight, and w3 is the temperature change weight, thus obtaining the optimization objective function.

[0135] Based on the optimization objective function, iterative calculations are performed, with the number of particles set to 50 and the maximum number of iterations to 500. The particles represent candidate solutions, the inertia weight is 0.8, and the learning factor is 2.0. The inertia weight indicates the degree to which the particle maintains its current velocity, and the learning factor indicates the degree to which the particle learns from the optimal solution, to obtain the optimal time solution.

[0136] The optimal time solution is reconstructed into a sequence, and a continuous time curve is generated through the cubic spline interpolation method. The continuous time curve is dynamically corrected, and real-time calibration and smoothing are performed according to temperature changes and state of charge changes to obtain the charge and discharge time prediction curve.

[0137] Specifically, the evidence reasoning analysis of the battery state evaluation index is carried out to calculate the charge and discharge time cohesion factor The calculation formula is:

[0138] ;

[0139] in, Represents the cohesion factor during charge and discharge time, is the weight coefficient of charging time and discharging time, and The reliability of the battery's time parameters during the charge and discharge process. Reliability is calculated by combining historical data with real-time monitoring data, for example, based on the battery status assessment results for each time period. Dynamic boundary constraints are calculated using the time characteristic correlation coefficient. The minimum and maximum charge times are determined using the following two functions:

[0140] ;

[0141] ;

[0142] in, is the minimum charging time, is the maximum charging time, and Represent the initial minimum and maximum charging time values, is the number of cycles of the battery, and and is the attenuation coefficient of the charging time, obtained through data fitting. These functions reflect the changing trend of charging time with the number of cycles, helping to optimize the charging process. The charging time constraint is mapped to the discharge time. Based on the relationship between battery capacity and discharge rate, a discharge time constraint function is constructed:

[0143] ;

[0144] ;

[0145] in, and are the minimum and maximum discharge times, and are the corresponding minimum and maximum discharge time coefficients. These two constraint functions ensure the time relationship of the charging and discharging process, so that the discharge time and the charging time maintain a reasonable ratio. The complete time constraint conditions are input into the time optimizer. In this process, the number of discrete points of the charging time and the discharging time is set to 100, and these discrete points are used to represent the time sampling points in the charging and discharging process. In order to optimize the charging and discharging time, the particle swarm optimization algorithm is used for solution. The algorithm finds the optimal solution by simulating the movement of particles in the search space. When constructing the time optimization objective function, its form is set as:

[0146] ;

[0147] in, is the objective function value, It is the charging time, is the discharge time, is the temperature change during the charge and discharge process, 、 、 Representing the weights of charge time, discharge time, and temperature variation, respectively. The objective function comprehensively considers the impact of various factors on the charge and discharge process, making the optimization process more comprehensive. During the iterative calculation, the number of particles was set to 50, and the maximum number of iterations was 500. During this process, each particle represents a candidate solution, with an inertia weight of 0.8 and a learning factor of 2.0. The inertia weight is used to maintain the particle's current velocity, while the learning factor controls the particle's ability to learn toward the optimal solution. Through multiple iterative calculations, the optimal time solution is obtained. The optimal time solution is then sequence-reconstructed. A continuous time curve is generated using the cubic spline interpolation method, which effectively smooths time series data. The generated continuous time curve reflects the changing trends during the charge and discharge process. The continuous time curve is dynamically corrected, calibrated and smoothed based on real-time temperature and state-of-charge changes. For example, if a sudden temperature increase is detected during a charging phase, the charging strategy is modified to adjust the charging time to ensure that the battery operates within a safe temperature range. Through real-time calibration and correction, a predicted charge and discharge time curve is ultimately obtained.

[0148] In a specific embodiment, the process of executing step S600 may specifically include the following steps:

[0149] Perform time segment control on the charge and discharge time prediction curve, set the control interval according to the charge time range and discharge time range, and obtain a segment control sequence;

[0150] Analyze the charge and discharge mode switching strategy, extract the mode switching criteria and control parameters corresponding to the current working condition from the strategy execution table, and obtain the mode control instructions;

[0151] According to the segmented control sequence and mode control instructions, the parameters are matched, and the charging current value, charging cut-off voltage value and discharge cut-off voltage value are configured online to obtain real-time control parameters;

[0152] Perform security constraints on real-time control parameters to obtain safety control parameters;

[0153] A feedback controller is constructed based on the safety control parameters, and a PID control algorithm is used to perform closed-loop control of the charging current, where the proportional coefficient Kp is 0.8, the integral coefficient Ki is 0.1, and the differential coefficient Kd is 0.05 to obtain the control output signal;

[0154] The control output signal is converted into power, and the on-time of the power tube is controlled by PWM modulation to obtain the charge and discharge execution instructions.

[0155] Specifically, for the time segment control of the charge and discharge time prediction curve, the corresponding control interval is set according to the charging time range and the discharging time range. Assume that the charging time range is arrive , the discharge time range is arrive , the control interval is expressed as:

[0156] ;

[0157] ;

[0158] By analyzing the charge and discharge time prediction curve, it is divided into time segments to obtain a segmented control sequence , where each segment corresponds to a different time interval and is dynamically adjusted according to the battery status. The segmented control sequence can effectively correspond to different charge and discharge stages, ensuring that the battery operates in the best state. When analyzing the charge and discharge mode switching strategy, the mode switching criteria and control parameters corresponding to the current working condition are extracted from the strategy execution table to form a mode control instruction. Set the mode switching criteria to , whose value is obtained by real-time monitoring of the battery status, such as battery temperature, charging current, discharging current, etc. When certain conditions are met, the trigger mode is switched, and the control parameters include the charging current value , charging cut-off voltage value and discharge cut-off voltage When matching parameters based on segment control sequences and mode control instructions, the following formula is used:

[0159] ;

[0160] ;

[0161] ;

[0162] in 、 、 It is a function that is dynamically configured according to the charge and discharge status. It can adjust the charge and discharge parameters in real time according to the segmented control sequence and the current mode control instructions. The real-time control parameters configured online can improve the charge and discharge efficiency and safety of the battery. After the real-time control parameters are safety constrained, the safety control parameters are obtained. Safety constraints are implemented by setting thresholds, for example, maximum and minimum values for charging current and voltage:

[0163] ;

[0164] ;

[0165] in, and These are the maximum allowable charging current and charging voltage, respectively. This ensures battery safety during charging and discharging, preventing overcharging or overdischarging. The core of building a feedback controller is the PID control algorithm. The basic formula for a PID controller is:

[0166] ;

[0167] in, is the control output signal, is the deviation of the system, calculated as the difference between the expected value and the actual value, is the proportionality coefficient, is the integration coefficient, is the differential coefficient. Set to 0.8, the integral coefficient Set to 0.1, the differential coefficient Set to 0.05, the selection of these coefficients is obtained through experiments and tuning, the purpose is to achieve fast response and good stability. The control output signal obtained by calculation It is used to control the change of charging current and perform power conversion based on this signal. During the power conversion process, pulse width modulation (PWM) technology is used to control the on-time of the power tube to accurately adjust the charging current. The basic formula of PWM modulation is:

[0168] ;

[0169] in, is the time when the power tube is turned on. This is the time the power transistor is off. By adjusting the ratio of on-time to off-time, the current during the charge and discharge process is precisely controlled, achieving effective battery management. The charge and discharge execution instructions generated through the above steps guide the battery management system's charge and discharge operations.

[0170] See also Figure 2 , Figure 2 A schematic block diagram of the structure of the performance optimization test system 200 for new energy batteries provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the new energy battery performance optimization test system 200 includes:

[0171] The acquisition module 210 is used to collect multi-channel operating status data of the new energy battery, and perform wavelet transform filtering and normalization processing to obtain a standardized feature data matrix;

[0172] Mode analysis module 220, used to perform operation mode analysis on the standardized characteristic data matrix and obtain a charge-discharge mode switching strategy through operating condition determination and calculation;

[0173] Extraction module 230, for inputting the standardized feature data matrix into a multi-branch feature extraction network for feature extraction to obtain a parameter-related feature vector;

[0174] A correlation analysis module 240 is used to perform grey correlation analysis on the parameter correlation feature vector to obtain a battery state evaluation index;

[0175] The calculation module 250 is used to input the battery state evaluation index into the charge and discharge time optimization model, and obtain the charge and discharge time prediction curve based on the evidence reasoning rule and the cohesion factor calculation;

[0176] The control module 260 is used to control the charging and discharging process of the new energy battery in real time according to the charging and discharging time prediction curve and the charging and discharging mode switching strategy.

[0177] Through the collaborative cooperation of the above components, multi-channel data acquisition and wavelet transform filtering processing are used to effectively improve the quality of the original data and reduce noise interference. The multi-branch feature extraction network architecture is used to realize deep feature extraction of three types of data: charge and discharge characteristics, temperature characteristics, and life characteristics, thereby improving the feature expression ability and enhancing the adaptability of the model to different working conditions. Based on grey correlation analysis and fuzzy comprehensive evaluation methods, a complete battery status assessment system is constructed, which improves the accuracy and reliability of status assessment. Through evidence reasoning rules and cohesion factor calculation, a charge and discharge time optimization model is established, which realizes accurate prediction of charge and discharge time and improves the efficiency of the charge and discharge process. Combined with dynamic boundary constraints and PID control algorithms, real-time and precise control of the charge and discharge process is realized, ensuring the safety and stability of the charge and discharge process. A multi-mode switching strategy is adopted to adaptively adjust control parameters according to different working conditions, thereby improving the flexibility and intelligence level of the battery management system.

[0178] See also Figure 3 , Figure 3 This is a schematic block diagram of the structure of a performance optimization test device 300 for a new energy battery provided in an embodiment of the present application. The performance optimization test device 300 for a new energy battery includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected via a system bus 303, wherein the memory 302 may include a non-volatile storage medium and an internal memory.

[0179] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 301, the processor 301 can execute any of the above-mentioned new energy battery performance optimization test methods.

[0180] The processor 301 is used to provide computing and control capabilities to support the operation of the entire new energy battery performance optimization test equipment 300.

[0181] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can execute any of the above-mentioned performance optimization test methods for new energy batteries.

[0182] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the performance optimization test equipment 300 of the new energy battery involved in the solution of the present application. The specific performance optimization test equipment 300 of the new energy battery may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0183] It should be understood that the processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0184] It should be noted that technical personnel in the relevant field can clearly understand that for the convenience and conciseness of description, the specific working process of the performance optimization test equipment 300 of the new energy battery described above can refer to the corresponding process of the aforementioned performance optimization test method for new energy batteries, and will not be repeated here.

[0185] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the one or more processors implement the performance optimization test method for new energy batteries provided in the embodiment of the present application.

[0186] The computer-readable storage medium may be an internal storage unit of the performance optimization test device 300 of the new energy battery in the aforementioned embodiment, such as a hard disk or memory of the performance optimization test device 300 of the new energy battery. The computer-readable storage medium may also be an external storage device of the performance optimization test device 300 of the new energy battery, such as a plug-in hard disk, a smart memory card (SMC), a secure digital (SD) card, a flash card, etc. equipped with the performance optimization test device 300 of the new energy battery.

[0187] Those skilled in the art will 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.

[0188] 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, 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. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute 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 code.

[0189] 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 above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above 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 performance optimization test method for new energy batteries, characterized in that: include: Collect multi-channel operating status data of new energy batteries, perform wavelet transform filtering and standardization processing, and obtain a standardized feature data matrix; Performing an operating mode analysis on the standardized characteristic data matrix, and obtaining a charge-discharge mode switching strategy through operating condition determination and calculation; Inputting the standardized feature data matrix into a multi-branch feature extraction network to perform feature extraction to obtain a parameter-associated feature vector; Performing grey correlation analysis on the parameter correlation feature vector to obtain a battery state evaluation index; Inputting the battery state evaluation index into the charge and discharge time optimization model, and calculating based on the evidence reasoning rule and the cohesion factor to obtain the charge and discharge time prediction curve; The charging and discharging process of the new energy battery is controlled in real time according to the charging and discharging time prediction curve and the charging and discharging mode switching strategy.

2. The performance optimization test method for new energy batteries according to claim 1, characterized in that: The multi-channel operating status data of the new energy battery is collected, and wavelet transform filtering and standardization processing are performed to obtain a standardized feature data matrix, including: The charging voltage, charging current, discharging voltage, and discharging current of new energy batteries are sampled at a frequency of 100 Hz, the surface temperature at a frequency of 10 Hz, the ambient temperature at a frequency of 1 Hz, and the internal resistance at a frequency of 0.1 Hz to obtain multi-channel operating status data; Transmitting the multi-channel operating status data to obtain a pre-processed data set, and performing wavelet transform processing on the pre-processed data set to obtain a noise reduction data set; Performing 3σ outlier detection on the denoised dataset to obtain a valid dataset, and performing standardization processing on the valid dataset to obtain a normalized dataset; Performing time series synchronization processing on the normalized data set to obtain a synchronized data set, and classifying and arranging the synchronized data set according to charge and discharge characteristics, temperature characteristics, and life characteristics to obtain a feature data set; A matrix reconstruction operation is performed on the characteristic data set, and the charge and discharge characteristic data, the temperature characteristic data, and the life characteristic data are rearranged according to a time sequence and a parameter correspondence relationship to obtain a standardized characteristic data matrix.

3. The performance optimization test method for new energy batteries according to claim 2, characterized in that: The operating mode analysis of the standardized characteristic data matrix is performed, and the charge-discharge mode switching strategy is obtained by operating condition determination and calculation, including: Performing charge and discharge parameter feature extraction on the standardized characteristic data matrix to obtain a standard charge and discharge mode parameter set, and performing threshold analysis on the temperature data in the standardized characteristic data matrix to obtain a temperature regulation mode parameter set; analyzing the number of cycles and capacity data in the standardized characteristic data matrix to obtain a life optimization mode parameter set; constructing a mode feature vector space according to the standard charge and discharge mode parameter set, the temperature adjustment mode parameter set, and the life optimization mode parameter set to obtain a multi-mode feature matrix; Performing an operating condition classification calculation on the multi-mode feature matrix, obtaining operating condition matching data based on the Euclidean distance between the current battery state parameter and the mode feature vector, performing mode analysis based on the operating condition matching data, selecting the operating mode with the highest matching degree as the optimal mode for the current operating condition, and obtaining a mode switching sequence; Based on the mode switching sequence, smooth transition processing is performed, continuity constraint calculation is performed on current and voltage parameters to obtain a transition control sequence, and the transition control sequence is reconstructed according to a timing relationship to generate a charge and discharge mode switching strategy, which includes a mode switching criterion and a strategy execution table of control parameters.

4. The performance optimization test method for new energy batteries according to claim 3, characterized in that: The step of inputting the standardized feature data matrix into a multi-branch feature extraction network for feature extraction to obtain a parameter-associated feature vector comprises: Performing data separation on the standardized characteristic data matrix, inputting the charge and discharge characteristic data into a first characteristic branch network, inputting the temperature characteristic data into a second characteristic branch network, and inputting the life characteristic data into a third characteristic branch network; Performing sliding window processing on the charge and discharge characteristic data of the first feature branch network, with the window length set to 1000 data points and the sliding step size set to 100 data points, and obtaining a charge and discharge characteristic matrix by calculating the mean, standard deviation, peak value, and valley value; Performing a fast Fourier transform on the temperature characteristic data of the second characteristic branch network, selecting an amplitude spectrum within a frequency band of 0-50 Hz, and obtaining a temperature characteristic matrix through spectrum peak detection and energy distribution calculation; Performing wavelet transform decomposition on the life characteristic data of the third characteristic branch network, selecting wavelet coefficients of 5 scales, and obtaining a life characteristic matrix through statistical moment and energy entropy calculation; Input the charge-discharge feature matrix into the first feature extraction layer, which includes three convolutional layers. Each convolutional layer uses a ReLU activation function and a batch normalization layer, and the convolution kernel sizes are 3×3, 5×5, and 7×7, respectively, to obtain charge-discharge depth features. Input the temperature feature matrix into the second feature extraction layer, which includes three one-dimensional convolutional layers, each of which is configured with a maximum pooling layer and a LeakyReLU activation function, and the pooling size is set to 2 to obtain temperature depth features; Inputting the life feature matrix into the third feature extraction layer, the third feature extraction layer includes two fully connected layers and one transformer encoder layer, the transformer encoder layer includes four attention heads, to obtain life depth features; Performing three-way self-attention calculation on the charge-discharge depth feature, the temperature depth feature, and the life depth feature, constructing a correlation relationship between the features through a multi-head attention mechanism, and obtaining a three-way attention weight matrix; The charge and discharge depth features, the temperature depth features, and the life depth features are weightedly fused according to the three-way attention weight matrix, and a fused feature tensor is obtained through residual connection and layer normalization processing. The fused feature tensor is then subjected to dimensionality reduction processing through a fully connected layer to obtain a parameter-associated feature vector.

5. The performance optimization test method for new energy batteries according to claim 4, characterized in that: The grey correlation analysis is performed on the parameter correlation feature vector to obtain a battery state evaluation index, including: The grey correlation degree is calculated for the parameter association feature vector, and the correlation coefficient between parameters is calculated based on the formula γ=(1+|ξ|) / (1+|ξ|+|Δi-Δj|), where γ is the correlation coefficient between parameters, ξ is the resolution coefficient, which is 0.5, Δi and Δj are the change rates of the i-th parameter and the j-th parameter, respectively, to obtain the correlation matrix; According to the correlation matrix, a temperature-depth state function S=α·exp(-β·T)+λ·DOD² is constructed, where S is the temperature-depth state function, T is the temperature value, DOD is the discharge depth, and α, β, and λ are state function coefficients; Performing eigenvalue decomposition on the correlation matrix, extracting the main influencing factors through principal component analysis, calculating the weight of each parameter based on the contribution rate, and obtaining a dynamic weight coefficient; Constructing a fuzzy evaluation matrix based on the state function coefficient and the dynamic weight coefficient, setting charging efficiency, discharge capacity, temperature uniformity, and cycle life as evaluation factors, and obtaining a fuzzy relationship matrix; Performing hierarchical analysis on the fuzzy relationship matrix, calculating the weight vector of each evaluation factor by establishing a judgment matrix, using consistency test to ensure the rationality of the weight, and obtaining a comprehensive weight vector; The comprehensive weight vector and the fuzzy relationship matrix are compounded and fuzzy synthesis is performed using a weighted average method. The evaluation level is determined by the maximum membership principle to obtain a fuzzy evaluation result, and a battery state evaluation index is calculated based on the fuzzy evaluation result.

6. The performance optimization test method for new energy batteries according to claim 5, characterized in that: The battery state evaluation index is input into the charge and discharge time optimization model, and a charge and discharge time prediction curve is obtained based on the evidence reasoning rule and the cohesion factor calculation, including: Performing evidence reasoning analysis on the battery state assessment index, constructing a charge and discharge time convergence factor by calculating the formula CF=Σ(wi·ri) / Σwi, where CF is the charge and discharge time convergence factor, wi is the weight coefficient of the charge time and discharge time, and ri is the reliability of the time parameter during the battery charge and discharge process, and obtaining the time characteristic correlation coefficient; The dynamic boundary constraint condition is calculated based on the time characteristic correlation coefficient, and the function Tchg is used to calculate the dynamic boundary constraint condition. min =T0·exp(-μt) to calculate the minimum charging time, through the function Tchg max =T1·exp(-νt) to calculate the maximum charging time, where Tchg min is the minimum charging time, Tchg max is the maximum charging time, T0 is the initial minimum charging time value, T1 is the initial maximum charging time value, t is the number of battery cycles, μ and ν are the charging time attenuation coefficients, and the charging time constraint function is obtained; The charge time constraint function is mapped to the discharge time, and the discharge time constraint function Tdis is constructed based on the relationship between battery capacity and discharge rate. min =k1·Tchg min and Tdis max =k2·Tchg max , where Tdis min is the minimum discharge time, Tdis max is the maximum discharge time, k1 is the minimum discharge time coefficient, k2 is the maximum discharge time coefficient, and the complete time constraint condition is obtained; The complete time constraint condition is input into the time optimizer, and the number of discrete points of charging time is set to 100, and the number of discrete points of discharging time is set to 100, where the discrete points represent the time sampling points in the charging and discharging process. The particle swarm optimization algorithm is used to solve the problem and obtain the initial time series; Dynamically optimize the initial time series to construct a time optimization objective function F=w1·Tchg+w2·Tdis+w3·ΔT, where F is the objective function value, Tchg is the charging time, Tdis is the discharging time, ΔT is the temperature change during the charging and discharging process, w1 is the charging time weight, w2 is the discharging time weight, and w3 is the temperature change weight, to obtain the optimization objective function; An iterative calculation is performed based on the optimization objective function, with the number of particles set to 50 and the maximum number of iterations set to 500, where the particles represent candidate solutions, the inertia weight is 0.8, and the learning factor is 2.

0. The inertia weight represents the degree to which the current velocity of the particle is maintained, and the learning factor represents the degree to which the particle learns from the optimal solution, to obtain the optimal time solution. The optimal time solution is reconstructed in sequence, and a continuous time curve is generated by a cubic spline interpolation method. The continuous time curve is dynamically corrected, and real-time calibration and smoothing are performed according to temperature changes and state of charge changes to obtain a charge and discharge time prediction curve.

7. The performance optimization test method for new energy batteries according to claim 6, characterized in that: The method of controlling the charge and discharge process of the new energy battery in real time according to the charge and discharge time prediction curve and the charge and discharge mode switching strategy includes: Performing time segment control on the charge and discharge time prediction curve, setting a control interval according to the charge time range and the discharge time range, and obtaining a segment control sequence; parsing the charge-discharge mode switching strategy, extracting the mode switching criteria and control parameters corresponding to the current working condition from the strategy execution table, and obtaining a mode control instruction; Perform parameter matching according to the segmented control sequence and the mode control instruction, configure the charging current value, the charging cut-off voltage value, and the discharging cut-off voltage value online, and obtain real-time control parameters; Performing security constraints on the real-time control parameters to obtain security control parameters; A feedback controller is constructed based on the safety control parameters, and a PID control algorithm is used to perform closed-loop control on the charging current, wherein the proportional coefficient Kp is 0.8, the integral coefficient Ki is 0.1, and the differential coefficient Kd is 0.05, to obtain a control output signal; The control output signal is subjected to power conversion, and the on-time of the power tube is controlled by PWM modulation to obtain a charge and discharge execution instruction.

8. A performance optimization test system for new energy batteries, characterized in that: A method for performing a performance optimization test of a new energy battery according to any one of claims 1 to 7, comprising: The acquisition module is used to collect multi-channel operating status data of new energy batteries, and perform wavelet transform filtering and standardization processing to obtain a standardized feature data matrix; A mode analysis module is used to perform an operation mode analysis on the standardized characteristic data matrix and obtain a charge-discharge mode switching strategy through operating condition determination and calculation; An extraction module, configured to input the standardized feature data matrix into a multi-branch feature extraction network for feature extraction to obtain a parameter-associated feature vector; A correlation analysis module is used to perform grey correlation analysis on the parameter correlation feature vector to obtain a battery state evaluation index; A calculation module, configured to input the battery state evaluation index into a charge and discharge time optimization model, and calculate based on evidence reasoning rules and cohesion factors to obtain a charge and discharge time prediction curve; A control module is used to control the charging and discharging process of the new energy battery in real time according to the charging and discharging time prediction curve and the charging and discharging mode switching strategy.

9. A performance optimization test device for new energy batteries, characterized in that: The performance optimization test device for new energy batteries includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the performance optimization test device for the new energy battery to execute the performance optimization test method for the new energy battery according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the performance optimization test method for the new energy battery according to any one of claims 1 to 7 is implemented.

Images