Lithium battery health state estimation method and system based on partial charging curve reconstruction

Abstract

The invention discloses a lithium battery health state estimation method and system based on partial charging curve reconstruction, and the method comprises the steps: carrying out the preprocessing and dividing of a lithium battery charging and discharging data set, obtaining a partial charging segment data set, training a partial charging data reconstruction model through the partial charging segment data set, and obtaining a partial charging data reconstruction model; and inputting the partial charging curve in the low correlation range into a partial charging data reconstruction model, reconstructing the partial charging curve into a high correlation curve, extracting health feature information, and inputting the extracted health feature information into a lithium battery health state estimation model based on Gaussian process regression for model training and testing. According to the method, the trained lithium battery health state estimation model is obtained, real-time and accurate evaluation can be provided for the health state of the lithium battery only by using charging fragment data with a relatively small length based on the trained partial charging data reconstruction model and the lithium battery health state estimation model, and a guarantee is provided for safe, stable and reliable operation of the lithium battery. The invention can be applied to the technical field of lithium batteries.

Description

Technical Field

[0001] The present invention relates to the technical field of lithium batteries, and in particular to a method and system for estimating the health status of a lithium battery based on partial charging curve reconstruction. Background Art

[0002] In recent years, lithium-ion batteries have become one of the core technologies for electrochemical conversion and storage. Due to its many advantages such as high energy density, long cycle life, no memory effect, low self-discharge rate and environmental protection, it has been widely used in mobile electronic devices, electric vehicles, aerospace, and energy storage. However, with the increase of working time and cycle time, the active materials inside lithium-ion batteries gradually degrade due to the deterioration of internal structure and electrochemical side reactions, resulting in increased resistance and reduced capacity. State of Health (SOH) is a key indicator reflecting the degree of battery degradation, which is usually defined as the ratio of the current maximum capacity to the nominal capacity. When the SOH value of a battery is too low, its capacity and power performance decrease significantly, which may lead to battery failure or even catastrophic accidents. Therefore, accurate estimation of the battery's SOH is crucial to ensure the safe and efficient operation of lithium-ion batteries. In recent years, some researchers have proposed many SOH estimation methods, which can generally be divided into model-based methods and data-driven methods. Model-based methods mainly include electrochemical models, equivalent circuit models, and empirical models. These methods mainly involve in-depth analysis of the physical and chemical reaction characteristics inside the battery, as well as the variation patterns of parameters such as terminal voltage and internal resistance, to build a battery model that accurately describes the battery degradation and failure mechanism. However, model-based SOH estimation methods have encountered challenges in practical applications, including the difficulty of modeling the complex physical failure mechanisms during battery degradation and identifying model parameters.

[0003] Data-driven methods have become the mainstream method for SOH estimation due to their superior flexibility and adaptability. These methods do not rely on complex battery models, but directly learn the patterns and trends of battery aging and performance degradation from battery data. Most existing methods are mainly based on battery data collected during the charge and discharge process under constant conditions in a laboratory setting. However, in actual application scenarios, the initial and final state of charge (SOC) of lithium batteries during the charge and discharge cycle are often random, making it difficult to obtain complete charging process data. Therefore, it is of great practical significance to study SOH estimation methods using partial charging data. Summary of the invention

[0004] The purpose of the present invention is to solve one of the technical problems existing in the prior art to at least a certain extent.

[0005] To this end, one purpose of an embodiment of the present invention is to provide a method for estimating the health status of a lithium battery based on reconstruction of a partial charging curve. This method only requires the use of charging segment data of a shorter length to provide real-time and accurate assessment of the health status of the lithium battery, thereby providing a guarantee for the safe, stable and reliable operation of the lithium battery.

[0006] Another object of an embodiment of the present invention is to provide a lithium battery health status estimation system based on partial charging curve reconstruction.

[0007] In order to achieve the above technical objectives, the technical solutions adopted by the embodiments of the present invention include:

[0008] In a first aspect, an embodiment of the present invention provides a method for estimating the health status of a lithium battery based on partial charging curve reconstruction, comprising the following steps:

[0009] Acquire a preset charge and discharge data set, and perform data preprocessing on the charge and discharge data set to obtain a preprocessed data set;

[0010] Dividing the preprocessed data according to the voltage range to obtain a partial charging segment data set, and randomly dividing the partial charging segment data set to obtain a first training set and a first test set;

[0011] Constructing a partial charging data reconstruction model based on a convolutional neural network, inputting the first training set into the partial charging data reconstruction model for training, and testing it with the first test set to obtain the trained partial charging data reconstruction model;

[0012] Inputting the partial charging curve in the preset low correlation range into the trained partial charging data reconstruction model to obtain a reconstructed high correlation curve;

[0013] Performing feature extraction on the high correlation curve to obtain a health feature information set characterizing the health status of the lithium battery, and randomly dividing the health feature information set to obtain a second training set and a second test set;

[0014] Constructing a lithium battery health state estimation model based on Gaussian process regression, inputting the second training set into the lithium battery health state estimation model for training, and testing it with the second test set to obtain the trained lithium battery health state estimation model;

[0015] Obtain a target partial charging curve of a target lithium battery, reconstruct the target partial charging curve according to the trained partial charging data reconstruction model to obtain a target high correlation curve, input the target high correlation curve into the trained lithium battery health status estimation model to obtain a health status estimation result of the target lithium battery.

[0016] Further, in one embodiment of the present invention, the data preprocessing is performed on the charge and discharge data set to obtain a preprocessed data set, which specifically includes:

[0017] Performing data cleaning on the charge-discharge data set to obtain the cleaned charge-discharge data set;

[0018] The cleaned charge and discharge data set is smoothed to obtain the pre-processed data set.

[0019] Further, in one embodiment of the present invention, the pre-processed data is divided according to the voltage range to obtain a partial charging segment data set, which specifically includes:

[0020] Calculating the charging current-time curve in the charging and discharging data set to obtain a charging capacity-time curve;

[0021] Calculating the charging capacity-time curve and the charging voltage-time curve to obtain a first charging curve;

[0022] Performing interpolation processing and moving average filtering on the first charging curve to obtain a smoothed second charging curve;

[0023] The second charging curve is divided into a plurality of segments according to the voltage range to obtain the partial charging segment data set.

[0024] Furthermore, in one embodiment of the present invention, the construction of a partial charging data reconstruction model based on a convolutional neural network specifically includes:

[0025] Construct convolutional layers based on one-dimensional convolutional neural networks;

[0026] Construct activation functions based on ReLU neural network layers;

[0027] Construct a pooling layer based on a one-dimensional maxpool neural network;

[0028] The partially charged data reconstruction model is constructed according to the convolutional layer, the activation function, the pooling layer and the fully connected layer.

[0029] Further, in one embodiment of the present invention, the inputting the first training set into the partial charging data reconstruction model for training, and testing it through the first test set to obtain the trained partial charging data reconstruction model specifically includes:

[0030] Inputting the first training set into the partial charging data reconstruction model to obtain a model output value;

[0031] The loss function is used to calculate the model output value and the true value of the training set to obtain an error value;

[0032] Optimizing the model parameters by back propagation algorithm and gradient descent method based on the error value, and returning the first training set to the partial charging data reconstruction model;

[0033] When the preset number of training times is reached, the first test set is input into the trained partial charging data reconstruction model for testing. If the test passes, the trained partial charging data reconstruction model is obtained.

[0034] Further, in one embodiment of the present invention, the feature extraction of the high correlation curve is performed to obtain a health feature information set characterizing the health status of the lithium battery, which specifically includes:

[0035] Determine the electric quantity value at each voltage according to the high correlation curve, and calculate the electric quantity statistical characteristics according to the electric quantity value;

[0036] The health characteristic information is obtained by splicing the power statistical characteristics;

[0037] The power statistics characteristics include the maximum power value, the minimum power value and the average power value of the high correlation curve.

[0038] Furthermore, in one embodiment of the present invention, the construction of a lithium battery health state estimation model based on Gaussian process regression specifically includes:

[0039] Construct a covariance function based on the radial basis kernel function;

[0040] Construct a joint Gaussian distribution based on the mean function and covariance function;

[0041] According to the covariance function and the joint Gaussian distribution, the lithium battery health state estimation model is constructed based on Gaussian process regression.

[0042] Further, in one embodiment of the present invention, the inputting the second training set into the lithium battery health state estimation model for training, and testing it with the second test set to obtain the trained lithium battery health state estimation model specifically includes:

[0043] Using the second training set to train the lithium battery health state estimation model, finding the optimal hyperparameter by maximizing the log-likelihood function, and obtaining the trained lithium battery health state estimation model;

[0044] The second test set is input into the trained lithium battery health state estimation model for testing. If the test passes, the trained lithium battery health state estimation model is obtained.

[0045] In a second aspect, an embodiment of the present invention provides a lithium battery health status estimation system based on partial charging curve reconstruction, comprising:

[0046] A preprocessing module, used to obtain a preset charge and discharge data set, and perform data preprocessing on the charge and discharge data set to obtain a preprocessed data set;

[0047] A segment data set division module, used for dividing the preprocessed data according to the voltage range to obtain a partial charging segment data set, and randomly dividing the partial charging segment data set to obtain a first training set and a first test set;

[0048] a partial charging data reconstruction model training module, used for constructing a partial charging data reconstruction model based on a convolutional neural network, inputting the first training set into the partial charging data reconstruction model for training, and testing it with the first test set to obtain the trained partial charging data reconstruction model;

[0049] A high correlation curve reconstruction module, used for inputting a partial charging curve in a preset low correlation range into the trained partial charging data reconstruction model to obtain a reconstructed high correlation curve;

[0050] A feature extraction module, used to extract features from the high correlation curve to obtain a health feature information set characterizing the health status of the lithium battery, and randomly divide the health feature information set to obtain a second training set and a second test set;

[0051] A lithium battery health state estimation model training module, used to construct a lithium battery health state estimation model based on Gaussian process regression, input the second training set into the lithium battery health state estimation model for training, and test it with the second test set to obtain the trained lithium battery health state estimation model;

[0052] A health state estimation module is used to obtain a target partial charging curve of a target lithium battery, reconstruct the target partial charging curve according to a trained partial charging data reconstruction model to obtain a target high correlation curve, and input the target high correlation curve into the trained lithium battery health state estimation model to obtain a health state estimation result of the target lithium battery.

[0053] In a third aspect, an embodiment of the present invention provides a lithium battery health status estimation device based on partial charging curve reconstruction, comprising:

[0054] at least one processor;

[0055] at least one memory for storing at least one program;

[0056] When the at least one program is executed by the at least one processor, the at least one processor implements the above-mentioned lithium battery health status estimation method based on partial charging curve reconstruction.

[0057] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores a program executable by a processor, and the program executable by the processor is used to execute the above-mentioned method for estimating the health status of a lithium battery based on reconstruction of a partial charging curve when executed by the processor.

[0058] The advantages and beneficial effects of the present invention will be partly given in the following description, partly become apparent from the following description, or be understood through the practice of the present invention:

[0059] The embodiment of the present invention preprocesses and divides the lithium battery charge and discharge data set to obtain a partial charging segment data set, uses the partial charging segment data set to train a partial charging data reconstruction model, inputs the partial charging curve in the low correlation range into the partial charging data reconstruction model, reconstructs it into a high correlation curve and extracts health feature information, and inputs the extracted health feature information into a lithium battery health status estimation model based on Gaussian process regression for model training and testing to obtain a trained lithium battery health status estimation model. Based on the trained partial charging data reconstruction model and the lithium battery health status estimation model, only a shorter length of charging segment data is needed to provide real-time and accurate evaluation of the health status of the lithium battery, thereby providing a guarantee for the safe, stable and reliable operation of the lithium battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solution in the embodiments of the present invention, the following introduction is made to the drawings required for use in the embodiments of the present invention. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solution of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0061] Figure 1 A flowchart of a method for estimating the health status of a lithium battery based on partial charging curve reconstruction provided by an embodiment of the present invention;

[0062] Figure 2 A schematic diagram of the structure of a partial charging data reconstruction model provided by an embodiment of the present invention;

[0063] Figure 3A schematic diagram of a partial charging curve and a reconstructed high correlation curve provided by an embodiment of the present invention;

[0064] Figure 4 A schematic diagram of the training process of the lithium battery health status estimation model provided by an embodiment of the present invention;

[0065] Figure 5 A schematic diagram of the correlation between health characteristics and battery health status before and after reconstruction provided by an embodiment of the present invention;

[0066] FIG6(a), FIG6(b), FIG6(c) and FIG6(d) are schematic diagrams of health status estimation results provided by an embodiment of the present invention;

[0067] Figure 7 A structural block diagram of a lithium battery health status estimation system based on partial charging curve reconstruction provided by an embodiment of the present invention;

[0068] Figure 8 A structural block diagram of a lithium battery health status estimation device based on partial charging curve reconstruction provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0069] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and are not to be construed as limitations of the present invention. For the step numbers in the following embodiments, they are only provided for the convenience of explanation, and the order between the steps is not limited in any way, and the execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0070] In the description of the present invention, the meaning of "a plurality" is two or more than two. If there is a description of "a first" or "a second", it is only used for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features. In addition, unless otherwise defined, all technical and scientific terms used in this document have the same meaning as those commonly understood by those skilled in the art.

[0071] Reference Figure 1 The embodiment of the present invention provides a method for estimating the health status of a lithium battery based on partial charging curve reconstruction, which specifically includes the following steps:

[0072] S101, obtaining a preset charge and discharge data set, and performing data preprocessing on the charge and discharge data set to obtain a preprocessed data set;

[0073] S102, dividing the preprocessed data according to the voltage range to obtain a partial charging segment data set, and randomly dividing the partial charging segment data set to obtain a first training set and a first test set;

[0074] S103, constructing a partial charging data reconstruction model based on a convolutional neural network, inputting the first training set into the partial charging data reconstruction model for training, and testing it with the first test set to obtain a trained partial charging data reconstruction model;

[0075] S104, inputting the partial charging curve in the preset low correlation range into the trained partial charging data reconstruction model to obtain a reconstructed high correlation curve;

[0076] S105, extracting features from the high correlation curve to obtain a health feature information set characterizing the health status of the lithium battery, and randomly dividing the health feature information set to obtain a second training set and a second test set;

[0077] S106, constructing a lithium battery health state estimation model based on Gaussian process regression, inputting the second training set into the lithium battery health state estimation model for training, and testing it with the second test set to obtain a trained lithium battery health state estimation model;

[0078] S107, obtaining a target partial charging curve of a target lithium battery, reconstructing the target partial charging curve according to a trained partial charging data reconstruction model to obtain a target high correlation curve, inputting the target high correlation curve into a trained lithium battery health status estimation model to obtain a health status estimation result of the target lithium battery.

[0079] As an optional implementation, data preprocessing is performed on the charge and discharge data set to obtain a preprocessed data set, which specifically includes:

[0080] S1011, performing data cleaning on the charge and discharge data set to obtain a cleaned charge and discharge data set;

[0081] S1012: Smoothing the cleaned charge and discharge data set to obtain a preprocessed data set.

[0082] Specifically, a certain number of samples are selected from the mass-produced lithium battery packs as test objects, and the selected battery packs are subjected to initial performance tests, including capacity, internal resistance, voltage, etc., to ensure that the samples meet the test standards. Then, the performance degradation of lithium batteries in long-term use is simulated through repeated charge and discharge cycles, and the charging current, charging voltage, temperature, time and other data at each moment in the lithium battery charging cycle are collected in real time using BMS sensors or monitoring equipment as the data of the charging cycle, and the data of multiple consecutive charging cycles are spliced ​​together as the original data.

[0083] Data preprocessing of the charge and discharge data set is a key step to ensure the accuracy of subsequent data analysis, model training, and performance evaluation. It aims to extract, clean, transform, and normalize data from the original charge and discharge data set to generate a preprocessed data set. Export the original charge and discharge data from the battery test equipment or database, usually including parameters such as timestamp, voltage, current, temperature, and capacity, and use tools such as Python (such as the pandas library), R, or Excel to import the data into a computer for subsequent processing. Check whether there are missing values ​​or outliers in the data set (such as NaN, Inf, or values ​​that are significantly deviated from the normal range). For missing values, deletion, interpolation, or model-based prediction methods can be used to handle them; if there are duplicate records in the data set, they should be deleted to avoid interference with subsequent analysis; for outliers that are significantly deviated from the normal range, they can be identified and processed by statistical methods (such as the 3σ principle) or based on business logic. After completing data cleaning, save the cleaned data set in an appropriate format (such as CSV, Excel, etc.) for subsequent analysis.

[0084] As an optional implementation, the preprocessed data is divided according to the voltage range to obtain a partial charging segment data set, which specifically includes:

[0085] S1021, calculating the charging current-time curve in the charging and discharging data set to obtain a charging capacity-time curve;

[0086] S1022, calculating a charging capacity-time curve and a charging voltage-time curve to obtain a first charging curve;

[0087] S1023, performing interpolation processing and moving average filtering on the first charging curve to obtain a second charging curve after smoothing;

[0088] S1024: Divide the second charging curve into multiple segments according to the voltage range to obtain a partial charging segment data set.

[0089] Specifically, the method of dividing the partial charging segment data set is described below:

[0090] 1) Calculate the charging current-time curve in the charge and discharge data set to obtain the charging capacity-time curve.

[0091] The charging capacity-time curve reflects the relationship between charging capacity and time. There are obvious differences in the charging capacity-time curves under different charge and discharge cycles, which can characterize the degree of battery aging. The calculation expression is as follows:

[0092] Q k =∫0 k ηIdt

[0093] In the formula, Q k is the total amount of electricity charged up to time k, η is the charging efficiency, I is the charging current, and t is the charging time.

[0094] 2) Calculate the charging capacity-time curve and the charging voltage-time curve to obtain the charging curve (capacity-voltage).

[0095] The obtained charging capacity-time curve and charging voltage-time curve are matched one by one according to the recording time to obtain the charging capacity-voltage curve. In most practical applications, due to the stability and controllability of charging conditions, it is usually suitable to extract health features from the charging process.

[0096] 3) Perform interpolation processing and moving average filtering on the charging curve to obtain a smoothed charging curve.

[0097] During the process of collecting the original charge and discharge data of the battery by the sensor, due to the influence of physical accuracy limitations or measurement errors, the collected data may contain noise, which may have an adverse effect on the subsequent health status estimation, so the original charging curve needs to be interpolated and smoothed. The present invention adopts linear interpolation and moving average filtering method.

[0098] Linear interpolation is a mathematical interpolation method that uses a straight line connecting two known data points to determine the value of an unknown point between the two points. It is applicable to the case where the function is linear within a certain interval. In a specific embodiment of the present invention, the data set used is the CACLE lithium battery data set, and its voltage range is 2.7V to 4.2V, retaining 4 decimal places. Since the voltage difference between different adjacent sampling time points is often different, in order to make the charging curve close to a continuous curve, the charging curve is interpolated with ΔV=0.0001V as the interpolation gap.

[0099] Moving average filtering is a signal processing technology based on statistical laws. It regards continuous sampled data as a queue of fixed length, smoothes the signal by calculating the average value of the data in the queue, reduces random fluctuations and noise, and thus highlights the trends and laws in the data. It is suitable for stable or slowly changing signals. The moving average filtering formula is as follows:

[0100]

[0101] Where x is the original input, x move_avg is the processed output, and N is the smoothing window size. In a specific embodiment of the present invention, the smoothing window size is set to 30, that is, a moving average is performed on the interpolated charging data with a window of 0.003V.

[0102] 4) The smoothed charging curve is divided into multiple segments according to the voltage range to obtain a partial charging segment data set.

[0103] In order to address the challenges of randomness and fragmentation of battery data in practical applications, an appropriate voltage interval is selected to divide the entire charging cycle curve into multiple data segments of equal size. With an excessively large voltage interval, it is challenging to obtain a charging curve within the required range under actual operating conditions. In contrast, an excessively small voltage interval may be limited by the acquisition accuracy of the BMS. In addition, such a small voltage interval results in low SOH estimation accuracy due to insufficient information. Therefore, in this study, the voltage interval ΔV of the charging curve segment is set to 0.05V, ensuring that the charging curve segment remains narrow enough while remaining within the acquisition accuracy capability, thereby ensuring the accuracy of SOH estimation. In the later stages of battery aging, due to the decline in battery performance and capacity, the initial voltage of the battery charge increases with the increase in the number of cycles. At the same time, due to measurement errors, the final voltage fluctuates around the critical voltage. Therefore, a relatively reasonable voltage range is selected in a specific embodiment of the present invention to ensure that some charging fragments during the entire battery aging process can be used for reconstruction and SOH estimation. For the CALCE dataset, [3.75, 3.80] V, [3.80, 3.85] V, ..., [4.05, 4.10] V were selected as the low correlation range of the reconstruction model input, and the range of [4.10, 4.15] V was specified as the high correlation range of the output generated by the reconstruction model.

[0104] The specific embodiment of the present invention adopts the K-fold cross validation method, and divides the data into 4 parts of battery data according to the battery monomer, 3 of which are used as training sets, and the remaining 1 is used as a test set. In this way, 4 sets of training / test sets can be obtained, and 4 training and testing are performed. The final result is the average of the 4 test results, which can be used to judge the effectiveness and generalization ability of the model, and make necessary adjustments and improvements.

[0105] As an optional implementation, a partial charging data reconstruction model based on a convolutional neural network is constructed, which specifically includes:

[0106] S1031. Construct a convolutional layer based on a one-dimensional convolutional neural network;

[0107] S1032. Construct an activation function based on the ReLU neural network layer;

[0108] S1033, constructing a pooling layer based on a one-dimensional maxpool neural network;

[0109] S1034. Construct a partial charging data reconstruction model based on the convolution layer, activation function, pooling layer and fully connected layer.

[0110] Specifically, in practical applications, the charging and discharging conditions of electric vehicles are quite complex, making it challenging to obtain complete charging and discharging curves. Therefore, utilizing partial charging curves within a narrow voltage range is more suitable for real-time and fast SOH estimation in real scenarios. However, with the increase of battery aging cycles, the changing trend of partial charging curves in different voltage ranges is significant. If the voltage range of the input partial charging curve is relatively low, its downward trend with battery degradation is not so obvious, resulting in a significant decrease in the accuracy of SOH estimation. To address the above problems, a CNN-based partial charging data reconstruction model is adopted, which reconstructs the charging data from a low voltage range to a high voltage range, thereby enhancing the correlation between the extracted HFs and SOH. By capturing the local features of the input low-correlation range fragment data through convolution operations, the CNNs model is able to achieve accurate reconstruction.

[0111] like Figure 2 The figure shows a schematic diagram of the structure of the partially charged data reconstruction model provided by an embodiment of the present invention. The partially charged data reconstruction model based on CNN consists of three convolutional layers, a global pooling layer and a fully connected layer. In addition, after each convolutional layer, the rectified linear unit (ReLU) activation function is used to enhance the nonlinear fitting ability of the model and avoid overfitting. In this embodiment, the hyperparameters of the convolutional network layer are hidden channel number hidden_channels=32, and convolution kernel size kernel_size=3. The formula of the partially charged data reconstruction model based on CNN is expressed as follows:

[0112]

[0113] Among them, x c represents the input feature matrix of the convolutional layer, w (k) It is represented as the weight matrix of the kth convolution kernel, whose size is L; t is the convolution step; ReLU(·) is the activation function; Maxpool(·) is the maximum pooling layer, whose output is the maximum value of the input feature map elements.

[0114] As an optional implementation, the first training set is input into the partial charging data reconstruction model for training, and the first test set is used for testing to obtain a trained partial charging data reconstruction model, which specifically includes:

[0115] S1035, inputting the first training set into the partial charging data reconstruction model to obtain a model output value;

[0116] S1036, using the loss function to calculate the model output value and the true value of the training set to obtain an error value;

[0117] S1037, optimizing the model parameters through the back propagation algorithm and the gradient descent method based on the error value, and returning to input the first training set into the partial charging data reconstruction model;

[0118] S1038. When the preset number of training times is reached, the first test set is input into the trained partial charging data reconstruction model for testing. If the test passes, a trained partial charging data reconstruction model is obtained.

[0119] Specifically, the training set of the partial charging data set is input into the reconstruction model for training, the original charging curve is divided into multiple parts according to the voltage range, and the reconstruction model of the corresponding range is trained respectively using the partial charging curve in each voltage range. In the specific example of the present invention, the dimension of the input data set is cycle_size*feature_size, where cycle_size is the number of lithium battery charging cycles, and feature_size is the dimension of the input feature. The dimension of train_x is 2850*500, the dimension of train_y is 2850*500, the dimension of test_x is 1045*500, and the dimension of test_y is 1045*500. The performance of the reconstruction model is evaluated by K-fold cross validation, and the error between the model output value and the true value of the training set is calculated by the loss function. The loss function formula is as follows:

[0120] Root Mean Square Error:

[0121] Mean absolute error:

[0122] Mean absolute percentage error:

[0123] in is the predicted value of the reconstructed curve, y i is the true value of the charging curve, and N is the number of samples.

[0124] Based on the BP algorithm of information forward propagation and error back propagation, the weight parameters of the model are updated and trained, and the training optimization steps are repeated until the predetermined number of training times is reached to minimize the error, and the trained partial charging data reconstruction model is obtained, which can effectively reconstruct the charging curve in the low correlation range into the high correlation range. For the CALCE data set, [3.75, 3.80] V, [3.80, 3.85] V, ..., [4.05, 4.10] V are selected as the low correlation range of the reconstruction model input, and the range of [4.10, 4.15] V is specified as the high correlation range of the output generated by the reconstruction model. It is worth noting that in order to ensure the effectiveness of the reconstruction and reduce the complexity of model training, a separate reconstruction model is trained for each different low voltage range.

[0125] like Figure 3 The diagram shows a partial charging curve provided by an embodiment of the present invention and a schematic diagram of a reconstructed high correlation curve. By integrating information from a high correlation range, the reconstruction model can capture the potential pattern between a low correlation sequence and a high correlation sequence, thereby converting low voltage range data obtained from practical applications into information with a strong correlation to SOH. In a specific embodiment of the present invention, the original charging curve is divided into multiple parts according to the voltage range, and the partial charging curves in each voltage range are respectively input into the reconstruction model of the corresponding range, so that the partial charging curve is reconstructed from the low correlation range to the high correlation range. In this embodiment, voltage ranges such as [3.75V, 3.80V], [3.80V, 3.85V], [3.85V, 3.90V], [3.90V, 3.95V], [3.95V, 4.00V], [4.00V, 4.05V], and [4.05V, 4.10V] are low correlation ranges, and [4.10V, 4.15V] is a high correlation range.

[0126] As an optional implementation, feature extraction is performed on the high correlation curve to obtain a health feature information set characterizing the health status of the lithium battery, which specifically includes:

[0127] S1051, determining the electric quantity value at each voltage according to the high correlation curve, and calculating the electric quantity statistical characteristics according to the electric quantity value;

[0128] S1052, obtaining health characteristic information according to the statistical characteristics of electricity;

[0129] The power statistics characteristics include the maximum power value, the minimum power value and the average power value of the high correlation curve.

[0130] Specifically, the reconstructed charging curve is calculated to obtain its statistical features such as maximum value, minimum value and average value, and the statistical features are concatenated as input features. Figure 5 The figure shows a schematic diagram of the correlation between the health characteristics (maximum value, minimum value and average value) before and after reconstruction and the battery health state provided by an embodiment of the present invention. The upper part and the lower part represent the original health characteristics HFs extracted from the original curve and the reconstructed curve and the reconstructed health characteristics HFs, respectively. The results show that the reconstructed HFs has a stronger linear correlation with SOH than the original HFs. When the original voltage range is lower than 3.95V, the correlation between the reconstructed health characteristics and SOH is significantly improved. However, when the original voltage range exceeds 3.95V, the correlation coefficient is only slightly enhanced. The correlation between HFs and SOH is enhanced by the reconstruction model, thereby achieving a more accurate SOH estimation later.

[0131] The health feature information set is randomly divided to obtain a second training set and a second test set.

[0132] Specifically, this embodiment adopts the holdout method to randomly divide the data set into a training set and a test set in a ratio of 3:1, and then uses the training set to train the model. By evaluating the performance of the model on the test set, the effectiveness and generalization ability of the model can be judged, and necessary adjustments and improvements can be made.

[0133] As an optional implementation, a lithium battery health state estimation model based on Gaussian process regression is constructed, which specifically includes:

[0134] S1061. Constructing a covariance function based on a radial basis kernel function;

[0135] S1062. Constructing a joint Gaussian distribution based on a mean function and a covariance function;

[0136] S1063. According to the covariance function and the joint Gaussian distribution, a lithium battery health status estimation model is constructed based on Gaussian process regression.

[0137] Specifically, Gaussian process regression (GPR) is a probabilistic prediction method that treats both known and unknown data as a Gaussian process with probabilistic, non-parametric and flexible characteristics. A Gaussian process is a random process in which any finite set of random variables follows a Gaussian distribution. A random variable X obeys a one-dimensional Gaussian distribution, determined by its expectation (μ) and variance (σ). By extending the one-dimensional Gaussian distribution to a multivariate Gaussian distribution and then extending the multivariate Gaussian distribution to infinite dimensions, a random process f(x) that satisfies the Gaussian process can be obtained. The formula of the Gaussian process regression model is expressed as follows:

[0138]

[0139] where σ f is the standard deviation, l is the scaling scale of the input feature, and ε is the normal distribution N(0,σ 2 )’s Gaussian noise; x is the input feature of the training set, and y is the true SOH value; is the input feature of the test set, is the estimated SOH value. In particular, is generated according to a normal distribution, with a mean can be regarded as the estimated SOH value, and the covariance Can reflect the uncertainty of the estimate.

[0140] As an optional implementation, the second training set is input into the lithium battery health state estimation model for training, and the second test set is used for testing to obtain a trained lithium battery health state estimation model, which specifically includes:

[0141] S1064, using the second training set to train the lithium battery health state estimation model, and finding the optimal hyperparameter by maximizing the log-likelihood function to obtain the trained lithium battery health state estimation model;

[0142] S1065: Input the second test set into the trained lithium battery health state estimation model for testing. If the test passes, a trained lithium battery health state estimation model is obtained.

[0143] Specifically, Figure 4 The figure shows a schematic diagram of the training process of the lithium battery health state estimation model provided by an embodiment of the present invention. The training set of the health feature set is input into the lithium battery health state estimation model based on Gaussian process regression. The data dimension of the health feature set data set is cycle_size*feature_size, where cycle_size is the number of lithium battery charging cycles, and feature_size is the dimension of the input feature. The dimension of train_x is 2850*3, the dimension of tran_y is 2850*1, the dimension of test_x is 1045*3, and the dimension of test_y is 1045*1. The model training process consistent with S6 is adopted to finally obtain a lithium battery health state estimation model, which has good performance in estimating the battery health state on the test data set.

[0144] After 3000 training cycles, the test data set is input into the trained lithium battery health status prediction model for testing to obtain the prediction results. The root mean square error, mean square error, and mean absolute percentage error between the prediction results and the true values ​​are calculated as the performance indicators of the current model. The smaller the error, the better the model performance. The calculation formula of the performance indicator is as follows:

[0145]

[0146] The specific embodiment of the present invention uses voltage ranges such as [3.75, 3.80]V, [3.80, 3.85]V, ..., [4.05, 4.10]V as the low correlation range of the reconstruction model input, and specifies the voltage range of [4.10, 4.15]V as the high correlation range of the output generated by the reconstruction model. The specific test of the embodiment of the present invention is to select the optimizer Adam for model training, the learning rate lr = 0.001, and the number of iterations epoch = 3000. The estimation results of the partially charged data reconstruction model that was finally trained on the test set of the data set are as follows: Figure 5As shown in Figure 6, the Pearson correlation coefficients of the original health features and SOH are 0.9076, 0.7782 and 0.8848 respectively, and the Pearson correlation coefficients of the reconstructed health features and SOH are 0.9387, 0.9399 and 0.9392 respectively. The estimation results of the health state estimation model finally trained on the test set of the data set are shown in Figure 6(a), Figure 6(b), Figure 6(c) and Figure 6(d), where Figure 6(a) corresponds to the original charging curve in the range of [3.75, 3.80] V, Figure 6(b) corresponds to the original charging curve in the range of [3.85, 3.90] V, Figure 6(c) corresponds to the original charging curve in the range of [3.95, 4.00] V, and Figure 6(d) corresponds to the original charging curve in the range of [4.05, 4.10] V. When the original charging curve is in the range of [3.75, 3.80] V, most of the SOH estimation results are distributed near the reference line, and some points have obvious deviations; as the voltage range of the original charging curve increases, such as in the range of [4.05, 4.10] V, the SOH estimation results are mainly clustered near the reference line, with fewer deviation points and a lower degree of deviation; when the voltage range is lower than 3.9 V, the MAPE error is lower than 2.5%; and when the voltage exceeds 3.9 V, the MAPE error remains within 1%.

[0147] After obtaining the trained partial charging data reconstruction model and the lithium battery health status estimation model, first obtain the target partial charging curve of the target lithium battery to be estimated, reconstruct the target partial charging curve according to the trained partial charging data reconstruction model to obtain a target high correlation curve, and then input the target high correlation curve into the trained lithium battery health status estimation model, so as to obtain the health status estimation result of the target lithium battery.

[0148] The method steps of the embodiment of the present invention are described above. It can be recognized that the embodiment of the present invention preprocesses and divides the lithium battery charge and discharge data set to obtain a partial charging segment data set, uses the partial charging segment data set to train the partial charging data reconstruction model, inputs the partial charging curve in the low correlation range into the partial charging data reconstruction model, reconstructs it into a high correlation curve and extracts health feature information, and inputs the extracted health feature information into the lithium battery health state estimation model based on Gaussian process regression for model training and testing, and obtains a trained lithium battery health state estimation model. Based on the trained partial charging data reconstruction model and the lithium battery health state estimation model, only a short length of charging segment data is needed to provide a real-time and accurate evaluation of the health state of the lithium battery. The voltage range of these charging segments is not limited and can be flexibly adjusted within the working voltage range of the battery, so that it is suitable for practical applications, providing a guarantee for the stable, safe and reliable operation of the lithium battery.

[0149] Reference Figure 7 The embodiment of the present invention provides a lithium battery health status estimation system based on partial charging curve reconstruction, comprising:

[0150] A preprocessing module is used to obtain a preset charge and discharge data set, perform data preprocessing on the charge and discharge data set, and obtain a preprocessed data set;

[0151] A segment data set division module is used to divide the preprocessed data according to the voltage range to obtain a partial charging segment data set, and randomly divide the partial charging segment data set to obtain a first training set and a first test set;

[0152] A partial charging data reconstruction model training module is used to construct a partial charging data reconstruction model based on a convolutional neural network, input the first training set into the partial charging data reconstruction model for training, and test it through the first test set to obtain a trained partial charging data reconstruction model;

[0153] A high correlation curve reconstruction module is used to input a partial charging curve in a preset low correlation range into a trained partial charging data reconstruction model to obtain a reconstructed high correlation curve;

[0154] A feature extraction module is used to extract features from the high correlation curve to obtain a health feature information set characterizing the health status of the lithium battery, and randomly divide the health feature information set to obtain a second training set and a second test set;

[0155] A lithium battery health state estimation model training module is used to build a lithium battery health state estimation model based on Gaussian process regression, input the second training set into the lithium battery health state estimation model for training, and test it with the second test set to obtain a trained lithium battery health state estimation model;

[0156] The health state estimation module is used to obtain the target partial charging curve of the target lithium battery, reconstruct the target partial charging curve according to the trained partial charging data reconstruction model to obtain the target high correlation curve, and input the target high correlation curve into the trained lithium battery health state estimation model to obtain the health state estimation result of the target lithium battery.

[0157] The contents of the above method embodiments are all applicable to the present system embodiments. The functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0158] Reference Figure 8 The embodiment of the present invention provides a lithium battery health status estimation device based on partial charging curve reconstruction, comprising:

[0159] at least one processor;

[0160] at least one memory for storing at least one program;

[0161] When the at least one program is executed by the at least one processor, the at least one processor implements the lithium battery health status estimation method based on partial charging curve reconstruction.

[0162] The contents of the above method embodiments are all applicable to the present device embodiments. The functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0163] An embodiment of the present invention also provides a computer-readable storage medium, which stores a program executable by a processor. When the program executable by the processor is executed by the processor, it is used to execute the above-mentioned lithium battery health status estimation method based on partial charging curve reconstruction.

[0164] A computer-readable storage medium of an embodiment of the present invention can execute a lithium battery health status estimation method based on partial charging curve reconstruction provided by a method embodiment of the present invention, can execute any combination of implementation steps of the method embodiment, and has the corresponding functions and beneficial effects of the method.

[0165] The embodiment of the present invention also discloses a computer program product or a computer program, wherein the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes Figure 1 The method shown.

[0166] In some selectable embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the above-mentioned boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided by way of example, for the purpose of providing a more comprehensive understanding of technology. The disclosed method is not limited to the operation and logic flow presented herein. Selectable embodiments are expected, wherein the order of various operations is changed and the sub-operation of a part for which is described as a larger operation is performed independently.

[0167] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise specified, one or more of the above-mentioned functions and / or features can be integrated into a single physical device and / or software module, or one or more functions and / or features can be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present invention. More specifically, in view of the properties, functions and internal relationships of the various functional modules in the device disclosed herein, the actual implementation of the module will be understood within the conventional skills of the engineer. Therefore, those skilled in the art can implement the present invention set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0168] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the 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, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the above methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0169] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0170] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and editable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the above-mentioned program is printed, since the above-mentioned program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or processing in other suitable ways as necessary, and then stored in a computer memory.

[0171] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0172] In the above description of this specification, the description with reference to the terms "one embodiment / example", "another embodiment / example" or "certain embodiments / examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0173] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

[0174] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A method for estimating the health status of a lithium battery based on partial charging curve reconstruction, characterized in that: The following steps are involved: Acquire a preset charge and discharge data set, and perform data preprocessing on the charge and discharge data set to obtain a preprocessed data set; Dividing the preprocessed data according to the voltage range to obtain a partial charging segment data set, and randomly dividing the partial charging segment data set to obtain a first training set and a first test set; Constructing a partial charging data reconstruction model based on a convolutional neural network, inputting the first training set into the partial charging data reconstruction model for training, and testing it with the first test set to obtain the trained partial charging data reconstruction model; Inputting the partial charging curve in the preset low correlation range into the trained partial charging data reconstruction model to obtain a reconstructed high correlation curve; Performing feature extraction on the high correlation curve to obtain a health feature information set characterizing the health status of the lithium battery, and randomly dividing the health feature information set to obtain a second training set and a second test set; Constructing a lithium battery health state estimation model based on Gaussian process regression, inputting the second training set into the lithium battery health state estimation model for training, and testing it with the second test set to obtain the trained lithium battery health state estimation model; Obtain a target partial charging curve of a target lithium battery, reconstruct the target partial charging curve according to the trained partial charging data reconstruction model to obtain a target high correlation curve, input the target high correlation curve into the trained lithium battery health status estimation model to obtain a health status estimation result of the target lithium battery.

2. A lithium battery health status estimation method based on partial charging curve reconstruction according to claim 1, characterized in that: The step of performing data preprocessing on the charge and discharge data set to obtain a preprocessed data set specifically includes: Performing data cleaning on the charge-discharge data set to obtain the cleaned charge-discharge data set; The cleaned charge and discharge data set is smoothed to obtain the pre-processed data set.

3. The method for estimating the health status of a lithium battery based on partial charging curve reconstruction according to claim 1, characterized in that: The pre-processed data is divided according to the voltage range to obtain a partial charging segment data set, which specifically includes: Calculating the charging current-time curve in the charging and discharging data set to obtain a charging capacity-time curve; Calculating the charging capacity-time curve and the charging voltage-time curve to obtain a first charging curve; Performing interpolation processing and moving average filtering on the first charging curve to obtain a smoothed second charging curve; The second charging curve is divided into a plurality of segments according to the voltage range to obtain the partial charging segment data set.

4. The method for estimating the health status of a lithium battery based on partial charging curve reconstruction according to claim 1, characterized in that: The construction of a partial charging data reconstruction model based on a convolutional neural network specifically includes: Construct convolutional layers based on one-dimensional convolutional neural networks; Construct activation functions based on ReLU neural network layers; Construct a pooling layer based on a one-dimensional maxpool neural network; The partially charged data reconstruction model is constructed according to the convolutional layer, the activation function, the pooling layer and the fully connected layer.

5. The method for estimating the health status of a lithium battery based on partial charging curve reconstruction according to claim 1, characterized in that: The step of inputting the first training set into the partial charging data reconstruction model for training, and testing the model using the first test set to obtain the trained partial charging data reconstruction model specifically includes: Inputting the first training set into the partial charging data reconstruction model to obtain a model output value; The loss function is used to calculate the model output value and the true value of the training set to obtain an error value; Optimizing the model parameters by back propagation algorithm and gradient descent method based on the error value, and returning the first training set to the partial charging data reconstruction model; When the preset number of training times is reached, the first test set is input into the trained partial charging data reconstruction model for testing. If the test passes, the trained partial charging data reconstruction model is obtained.

6. The method for estimating the health status of a lithium battery based on partial charging curve reconstruction according to claim 1, characterized in that: The feature extraction of the high correlation curve is performed to obtain a health feature information set characterizing the health status of the lithium battery, which specifically includes: Determine the electric quantity value at each voltage according to the high correlation curve, and calculate the electric quantity statistical characteristics according to the electric quantity value; The health characteristic information is obtained by splicing the power statistical characteristics; The power statistics characteristics include the maximum power value, the minimum power value and the average power value of the high correlation curve.

7. The method for estimating the health status of a lithium battery based on partial charging curve reconstruction according to claim 1, characterized in that: The construction of a lithium battery health status estimation model based on Gaussian process regression specifically includes: Construct a covariance function based on the radial basis kernel function; Construct a joint Gaussian distribution based on the mean function and covariance function; According to the covariance function and the joint Gaussian distribution, the lithium battery health state estimation model is constructed based on Gaussian process regression.

8. The method for estimating the health status of a lithium battery based on partial charging curve reconstruction according to claim 1, characterized in that: The step of inputting the second training set into the lithium battery health state estimation model for training, and testing the model using the second test set to obtain the trained lithium battery health state estimation model specifically includes: Using the second training set to train the lithium battery health state estimation model, finding the optimal hyperparameter by maximizing the log-likelihood function, and obtaining the trained lithium battery health state estimation model; The second test set is input into the trained lithium battery health state estimation model for testing. If the test passes, the trained lithium battery health state estimation model is obtained.

9. A lithium battery health status estimation system based on partial charging curve reconstruction, characterized in that: include: A preprocessing module, used to obtain a preset charge and discharge data set, and perform data preprocessing on the charge and discharge data set to obtain a preprocessed data set; A segment data set division module, used for dividing the preprocessed data according to the voltage range to obtain a partial charging segment data set, and randomly dividing the partial charging segment data set to obtain a first training set and a first test set; a partial charging data reconstruction model training module, used for constructing a partial charging data reconstruction model based on a convolutional neural network, inputting the first training set into the partial charging data reconstruction model for training, and testing it with the first test set to obtain the trained partial charging data reconstruction model; A high correlation curve reconstruction module, used for inputting a partial charging curve in a preset low correlation range into the trained partial charging data reconstruction model to obtain a reconstructed high correlation curve; A feature extraction module, used to extract features from the high correlation curve to obtain a health feature information set characterizing the health status of the lithium battery, and randomly divide the health feature information set to obtain a second training set and a second test set; A lithium battery health state estimation model training module, used to construct a lithium battery health state estimation model based on Gaussian process regression, input the second training set into the lithium battery health state estimation model for training, and test it with the second test set to obtain the trained lithium battery health state estimation model; A health state estimation module is used to obtain a target partial charging curve of a target lithium battery, reconstruct the target partial charging curve according to a trained partial charging data reconstruction model to obtain a target high correlation curve, and input the target high correlation curve into the trained lithium battery health state estimation model to obtain a health state estimation result of the target lithium battery.

10. A computer-readable storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to execute a lithium battery health status estimation method based on partial charging curve reconstruction as described in any one of claims 1 to 8 when executed by the processor.

Images