Safety early warning method, device and equipment for lithium ion battery and storage medium

Abstract

The invention relates to a safety early warning method, device and equipment for a lithium ion battery and a storage medium. The method comprises the following steps: acquiring a to-be-measured multivariate signal of the lithium ion battery in an operation process; the multivariate signal to be detected is input into a lithium ion battery safety early warning model, a fault diagnosis result output by the lithium ion battery safety early warning model is obtained, and the lithium ion battery safety early warning model is a model generated after training of a training set. The training set is generated after target multivariate signals of the sample lithium ion battery in the operation process are processed by using an improved black-wing plinual optimization algorithm; and performing safety early warning on the lithium ion battery according to a fault diagnosis result. By adopting the method, the accuracy of safety early warning of the lithium ion battery can be improved.

Description

Technical Field

[0001] The present application relates to the field of power electronics technology, and particularly to a safety warning method, device, equipment and storage medium for a lithium-ion battery. Background Art

[0002] With new energy vehicles becoming increasingly popular among people, the electric vehicle industry has developed rapidly. Due to its high energy density, long life and other advantages, lithium-ion batteries are widely used as the power source of electric vehicles. Therefore, the research and development of safe and reliable lithium battery safety warning methods have received increasing attention.

[0003] In related technologies, the safety warning of lithium-ion batteries mainly includes model-based methods and data-driven methods, which all focus on studying the relationship between the voltage of lithium-ion batteries and fault characteristics, extracting fault-characterizing features from the voltage, and using machine learning algorithms to find the relationship between features and faults.

[0004] However, the safety warning of lithium-ion batteries in related technologies only focuses on voltage signals and does not consider the influence of other factors such as temperature, pressure, and internal resistance during the operation of electric vehicles on battery faults, resulting in low accuracy of lithium-ion battery safety warning. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a safety warning method, device, equipment and storage medium for a lithium-ion battery that can improve the accuracy of safety warning of lithium-ion batteries.

[0006] In a first aspect, the present application provides a safety warning method for a lithium-ion battery, including:

[0007] Obtain a plurality of signals to be measured during the operation of the lithium-ion battery;

[0008] Input the plurality of signals to be measured into a lithium-ion battery safety warning model, and obtain a fault diagnosis result output by the lithium-ion battery safety warning model. The lithium-ion battery safety warning model is a model generated after being trained by a training set, and the training set is generated by processing the target plurality of signals of a sample lithium-ion battery during operation using an improved black-winged kite optimization algorithm;

[0009] Perform safety warning on the lithium-ion battery according to the fault diagnosis result.

[0010] In one of the embodiments, before inputting the plurality of target signals into the lithium-ion battery safety warning model and obtaining the fault diagnosis result output by the lithium-ion battery safety warning model, the method further includes:

[0011] Collect the target multi-source signals during the operation of the sample lithium-ion battery;

[0012] Generate a training set and a test set for the lithium-ion battery safety warning model according to the target multi-source signals;

[0013] Use the training set to train the lithium-ion battery safety warning model;

[0014] Use the test set to test the trained lithium-ion battery safety warning model.

[0015] In one embodiment, generating the training set and the test set for the lithium-ion battery safety warning model according to the target multi-source signals includes:

[0016] Divide the target multi-source signals into multiple calculation windows;

[0017] Use the improved black-winged kite optimization algorithm to perform variational mode decomposition on the sub-signals in each calculation window to obtain multiple intrinsic mode function components of the target multi-source signals;

[0018] Calculate the distribution entropy of each intrinsic mode function component;

[0019] Concatenate the distribution entropies of the intrinsic mode function components of the same sample lithium-ion battery at the beginning and end according to the signal type to obtain the corresponding feature vector of the sample lithium-ion battery;

[0020] Determine the training set and the test set of the lithium-ion battery safety warning model according to the feature vector of the sample lithium-ion battery.

[0021] In one embodiment, determining the training set and the test set of the lithium-ion battery safety warning model according to the feature vector of the sample lithium-ion battery includes:

[0022] Generate a first candidate sample set according to the feature vector of the sample lithium-ion battery and the fault type label, and the first candidate sample set includes sample data pairs composed of the feature vector and different fault type labels;

[0023] Use the correlation coefficient method to perform feature screening on the sample data pairs in the first candidate sample set to obtain a second candidate sample set, and the second candidate sample set includes target sample data pairs in which the screened feature vectors are positively correlated with the fault type labels;

[0024] Divide the training set and the test set from the second candidate sample set.

[0025] In one embodiment, the method further includes:

[0026] An improved black-winged kite optimization algorithm is obtained by integrating a dynamic reverse learning strategy and a golden sine strategy to improve the initial black-winged kite optimization algorithm.

[0027] In one embodiment, collecting the target multi-signal of the lithium-ion battery during operation includes:

[0028] Under the charge and discharge state, collecting the operation data of each battery cell of the sample lithium-ion battery during operation, where the sample lithium-ion battery is a battery pack formed by connecting a normal battery and a faulty battery in series;

[0029] The operation data of each battery cell is combined to form the target multi-signal.

[0030] In one embodiment, using the training set to train the lithium-ion battery safety warning model includes:

[0031] Using the improved black-winged kite optimization algorithm to optimize the hyperparameters of the random forest model to obtain the optimal parameters of the random forest model;

[0032] Inputting the optimal parameters of the random forest model into the random forest model;

[0033] Using the training set to train the random forest model input with the optimal parameters to obtain the lithium-ion battery safety warning model.

[0034] In a second aspect, the present application also provides a safety warning device for a lithium-ion battery, including:

[0035] An acquisition module for acquiring the multi-signal to be measured of the lithium-ion battery during operation;

[0036] A diagnosis module for inputting the multi-signal to be measured into the lithium-ion battery safety warning model and obtaining the fault diagnosis result output by the lithium-ion battery safety warning model, where the lithium-ion battery safety warning model is a model generated after training with a training set, and the training set is generated by processing the target multi-signal of the sample lithium-ion battery during operation using the improved black-winged kite optimization algorithm;

[0037] A warning module for performing safety warning on the lithium-ion battery according to the fault diagnosis result.

[0038] In one embodiment, the safety warning device for the lithium-ion battery further includes:

[0039] A training module, configured to collect target multi-source signals during the operation of the sample lithium-ion battery; generate a training set and a test set for the lithium-ion battery safety warning model according to the target multi-source signals; use the training set to train the lithium-ion battery safety warning model; use the test set to test the trained lithium-ion battery safety warning model.

[0040] In one embodiment, the training module is further configured to divide the target multi-source signals into multiple calculation windows; use the improved black-winged kite optimization algorithm to perform variational mode decomposition on the sub-signals in each calculation window to obtain multiple intrinsic mode function components of the target multi-source signals; calculate the distribution entropy of each intrinsic mode function component; concatenate the distribution entropies of the intrinsic mode function components of the same sample lithium-ion battery at the beginning and end according to the signal type to obtain the corresponding feature vector of the sample lithium-ion battery; determine the training set and the test set of the lithium-ion battery safety warning model according to the feature vector of the sample lithium-ion battery.

[0041] In one embodiment, the training module is further configured to generate a first candidate sample set according to the feature vector of the sample lithium-ion battery and the fault type label, where the first candidate sample set includes sample data pairs composed of the feature vector and different fault type labels; use the correlation coefficient method to perform feature screening on the sample data pairs in the first candidate sample set to obtain a second candidate sample set, where the second candidate sample set includes target sample data pairs in which the screened feature vector is positively correlated with the fault type label; divide the training set and the test set from the second candidate sample set.

[0042] In one embodiment, the training module is further configured to improve the initial black-winged kite optimization algorithm by fusing a dynamic backpropagation learning strategy and a golden sine strategy to obtain an improved black-winged kite optimization algorithm.

[0043] In one embodiment, the training module is further configured to collect the operation data of each battery cell during the operation of the sample lithium-ion battery in the charge and discharge state, where the sample lithium-ion battery is a battery pack formed by connecting a normal battery and a faulty battery in series; form the target multi-source signals with the operation data of each battery cell.

[0044] In one embodiment, the diagnosis module is further configured to use the improved black-winged kite optimization algorithm to optimize the hyperparameters of the random forest model to obtain the optimal parameters of the random forest model; input the optimal parameters of the random forest model into the random forest model; use the training set to train the random forest model input with the optimal parameters to obtain the lithium-ion battery safety warning model.

[0045] In a third aspect, the present application further provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the safety warning method for the lithium-ion battery in the above first aspect is implemented.

[0046] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the safety warning method for the lithium-ion battery in the above first aspect is implemented.

[0047] In a fifth aspect, the present application further provides a computer program product, including a computer program. When the computer program is executed by a processor, the safety warning method for the lithium-ion battery in the above first aspect is implemented.

[0048] For the safety warning method, device, equipment and storage medium of the above lithium-ion battery, first, the multi-signal to be measured during the operation of the lithium-ion battery is acquired; subsequently, the multi-signal to be measured is input into the lithium-ion battery safety warning model, and the fault diagnosis result output by the lithium-ion battery safety warning model is obtained. The lithium-ion battery safety warning model is a model generated after being trained by a training set, and the training set is generated after processing the target multi-signal of the sample lithium-ion battery during operation using an improved black-winged kite optimization algorithm; finally, according to the fault diagnosis result, safety warning is performed on the lithium-ion battery. Since the lithium-ion battery safety warning model can extract the characteristics of the multi-signal during the operation of the lithium-ion battery, the diversity of the characteristics is improved. At the same time, the improved black-winged kite optimization algorithm is used to process the data in the training set of the lithium-ion battery safety warning model, so that more sufficient search can be performed around during the optimization process, avoiding falling into local optimal solutions, and thus the accuracy of the safety warning of the lithium-ion battery can be improved. Description of the Drawings

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other related drawings can be obtained based on these drawings.

[0050] Figure 1 It is an application environment diagram of a safety warning method for a lithium-ion battery provided by an embodiment of the present application;

[0051] Figure 2 It is a flow schematic diagram of a safety warning method for a lithium-ion battery provided by an embodiment of the present application;

[0052] Figure 3Schematic diagram of the distribution entropy value of the IMF component of the voltage signal provided by the embodiment of the present application;

[0053] Figure 4 Schematic diagram of optimizing the VMD algorithm by using an improved black-winged kite optimization algorithm provided by the embodiment of the present application;

[0054] Figure 5 Schematic flow diagram of a safety warning method for a lithium-ion battery provided by the embodiment of the present application;

[0055] Figure 6 Block diagram of the structure of a safety warning device for a lithium-ion battery provided by the embodiment of the present application;

[0056] Figure 7 Internal structure diagram of a computer device provided by the embodiment of the present application. Detailed implementation manners

[0057] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0058] The safety warning method for a lithium-ion battery provided by the embodiment of the present application can be applied to an application environment as Figure 1 shown. Among them, the terminal 102 communicates with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or can be placed in the cloud or other network servers.

[0059] The terminal 102 can collect the multi-signals to be measured of the lithium-ion battery in real time during operation. Subsequently, the server 104 obtains the multi-signals to be measured of the lithium-ion battery in operation from the terminal 102, inputs the multi-signals to be measured into the lithium-ion battery safety warning model, and obtains the fault diagnosis result output by the lithium-ion battery safety warning model. The lithium-ion battery safety warning model is a model generated after being trained by a training set, and the training set is generated by processing the target multi-signals of the sample lithium-ion battery in operation by using an improved black-winged kite optimization algorithm. Finally, the server 104 performs a safety warning on the lithium-ion battery according to the fault diagnosis result.

[0060] Among them, the terminal 102 can be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0061] In an exemplary embodiment, as Figure 2 shown, a safety warning method for a lithium-ion battery is provided. Taking the method applied to the Figure 1 server in it as an example, it includes S201 - S203:

[0062] S201. Obtain the multi-signals to be measured during the operation of the lithium-ion battery.

[0063] In this application, the terminal can collect the multi-signals to be measured during the operation of the lithium-ion battery in real time. Correspondingly, the server can obtain the multi-signals to be measured during the operation of the lithium-ion battery from the terminal, so as to perform subsequent fault diagnosis of the lithium-ion battery.

[0064] It should be understood that this application embodiment does not limit the multi-signals to be measured. In some embodiments, the multi-signals to be measured can include voltage signals, temperature signals, pressure signals, impedance signals, etc.

[0065] S202. Input the multi-signals to be measured into the lithium-ion battery safety warning model, and obtain the fault diagnosis result output by the lithium-ion battery safety warning model.

[0066] Among them, the above-mentioned lithium-ion battery safety warning model can be a random forest model optimized by an improved black-winged kite optimization algorithm for parameters.

[0067] In some embodiments, the fault diagnosis result output by the lithium-ion battery safety warning model can indicate whether the lithium-ion battery is faulty; if the fault diagnosis result can indicate that the lithium-ion battery to be measured is faulty, the fault diagnosis result can also include the fault type label of the lithium-ion battery.

[0068] In the embodiment of this application, the lithium-ion battery safety warning model is a model generated after being trained by a training set, and the training set is generated by processing the target multi-signals of the sample lithium-ion battery during operation using an improved black-winged kite optimization algorithm.

[0069] First, the improved Black Kite Optimization Algorithm will be described below.

[0070] It should be understood that the Black Kite Optimization Algorithm is a meta-heuristic optimization algorithm used for feature selection and model parameter optimization in machine learning. Correspondingly, the improved Black Kite Optimization Algorithm can be an algorithm obtained by optimizing and improving the Black Kite Optimization Algorithm through one or more strategies.

[0071] In some embodiments, the server can improve the initial Black Kite Optimization Algorithm by integrating the dynamic reverse learning strategy and the golden sine strategy to obtain the improved Black Kite Optimization Algorithm.

[0072] Among them, the dynamic reverse learning strategy is an adaptive learning method that can dynamically generate reverse solutions based on the information of the current solution to expand the search space and increase the possibility of the algorithm finding a better solution. The golden sine strategy is an optimization strategy that combines the golden ratio and the characteristics of the sine function and is used to solve optimization problems.

[0073] Exemplarily, in the process of generating the improved Black Kite Optimization Algorithm, the server can first generate the initial Black Kite Optimization Algorithm, that is, perform population initialization, create a matrix to represent the black kite population, and this matrix can be as shown in formula (1), and the initialization of the population allocation can be as shown in formula (2):

[0074]

[0075] X i = BK i,j + rand(BK ub - BK lb ) (2)

[0076] where p is the number of potential solutions, d is the dimension of the problem to be solved, BK is the solution, BK ub and BK lb are the boundaries of the solution, rand represents generating a random number, X i is the i-th population, and BK i,j is the j-th dimension of the i-th solution.

[0077] Exemplarily, in order to optimize the quality of the initial solution of the Black Kite Optimization Algorithm, the dynamic reverse learning strategy can be integrated during the population initialization process. During the process of integrating the dynamic reverse learning strategy, the individual with the best fitness can be selected as the initial leader. The calculation method can be as shown in formula (3):

[0078] X D = X i + r 1 ×(r 2 ×(BKlb +BK ub -X i )-X i ) (3)

[0079] Among them, X D is the newly generated population, r 1 and r 2 are random numbers uniformly distributed between 0 and 1.

[0080] Exemplarily, in order to improve the local optimization ability of the algorithm, the golden sine strategy can be integrated into the black-winged kite optimization algorithm for improvement. The improved mathematical model can be as shown in formula (4). Combining with the mathematical model of the migration behavior of the black-winged kite optimization algorithm shown in formula (5), the improved black-winged kite optimization algorithm can be obtained:

[0081]

[0082] Among them, is the value of the i-th black-winged kite in the j-th dimension at the k-th iteration, is the value of the i-th black-winged kite in the j-th dimension at the (k + 1)-th iteration; R 1 is a random number between [0, 2π], representing the distance that the black-winged kite moves; R 2 is a random number between [0, π], representing the direction that the black-winged kite moves; represents the optimal individual position at the k-th iteration; α 1 and α 2 are the golden section coefficients; S is the safety coefficient; γ is the iteration attenuation coefficient.

[0083] Among them, C is the Cauchy mutation, is the leading scorer of the j-th dimension of the black-winged kite at the k-th iteration, F i is the current position of any black-winged kite in the j-th dimension at the k-th iteration, F ri is the fitness value of any black-winged kite in the j-th dimension at the t-th iteration.

[0084] In this application, the black-winged kite optimization algorithm is improved, the initialization performance of the algorithm is improved, the solution is more fully searched around during the optimization process, getting stuck in local optimal solutions is avoided, and the convergence speed of the algorithm is increased, improving the performance of the algorithm.

[0085] The training process of the lithium-ion battery safety warning model will be described below.

[0086] In some embodiments, before inputting the target multi-source signal into the lithium-ion battery safety warning model and obtaining the fault diagnosis result output by the lithium-ion battery safety warning model, the server may first collect the target multi-source signal of the sample lithium-ion battery during operation. Secondly, the server may generate a training set and a test set for the lithium-ion battery safety warning model according to the target multi-source signal. Thirdly, the server may use the training set to train the lithium-ion battery safety warning model. Finally, the server may use the test set to test the trained lithium-ion battery safety warning model.

[0087] It should be understood that in some embodiments of the present application, regarding how to collect the target multi-source signal of the sample lithium-ion battery during operation, the server may collect the operation data of each battery cell of the sample lithium-ion battery during operation in the charge and discharge states, and the sample lithium-ion battery is a battery pack formed by connecting a normal battery and a faulty battery in series. Subsequently, the server may form the operation data of each battery cell into the target multi-source signal.

[0088] Among them, the above sample lithium-ion battery can be obtained by electrically abusing a normal lithium-ion battery through overcharging, over-discharging, and parallel resistance between the positive and negative electrodes, etc., to make it a faulty battery, and then using the battery pack formed by connecting the normal battery and the faulty battery in series as the sample lithium-ion battery.

[0089] It should be understood that the present application embodiment does not limit the charge and discharge states of the sample lithium-ion battery. Exemplarily, the sample lithium-ion battery can be charged at a constant current of 1C during the charging stage and discharged under any working conditions during the discharging stage.

[0090] Exemplarily, the target multi-source signal may include a voltage signal, a temperature signal, a pressure signal, and an impedance signal. After the server collects the voltage signal, temperature signal, pressure signal, and impedance signal of the sample lithium-ion battery during operation, it may form a data matrix as the target multi-source signal. The data matrix may be as shown in formula (6):

[0091]

[0092] Among them, v represents the voltage signal, t represents the temperature signal, f represents the pressure signal, r represents the impedance signal, and n represents the maximum sampling point.

[0093] It should be understood that the embodiments of the present application do not limit how to generate the training set and test set of the lithium-ion battery safety warning model according to the target multi-source signals. In some embodiments, the server may first divide the target multi-source signals into multiple calculation windows. Secondly, the server uses the improved black-winged kite optimization algorithm to perform variational mode decomposition on the sub-signals in each calculation window to obtain multiple intrinsic mode function components of the target multi-source signals. Thirdly, the server calculates the distribution entropy of each intrinsic mode function component and splices the distribution entropies of the intrinsic mode function components of the same sample lithium-ion battery at the head and tail according to the signal type to obtain the corresponding feature vector of the sample lithium-ion battery. Finally, the server determines the training set and test set of the lithium-ion battery safety warning model according to the feature vector of the sample lithium-ion battery.

[0094] Among them, the size of the above calculation window can be specifically set according to the actual situation. Exemplarily, a calculation window can be divided for the data matrix A formed by the target multi-source data. The window length can be set to m (1 < m < n). Each calculation window contains m rows and 4 columns of data. The calculation window slides backward by m / 2 data each time until the end of the data. If the remaining data at the end is less than m / 2, the remaining data is deleted.

[0095] In some optional embodiments, after dividing the target multi-source signals into multiple calculation windows, the sub-signals divided in the calculation window can also be normalized. Subsequently, variational mode decomposition (VMD) can be performed column by column. During the variational mode decomposition process, the improved black-winged kite optimization algorithm (MBKA) can be used to optimize the decomposition mode number and penalty factor in the VMD algorithm, and the data in the calculation window is decomposed into multiple intrinsic mode function (IMF) components and residuals.

[0096] Exemplarily, for each IMF component among all the decomposed IMF components, the distribution entropy of each IMF component can be calculated by formula (7).

[0097]

[0098] Among them, DE is the distribution entropy of the IMF component, H is the maximum value of the number of groups of the distribution entropy histogram, is the empirical probability density of the IMF component under the corresponding histogram group number h.

[0099] Exemplarily, after calculating the distribution entropy of each IMF component, all the IMF components can be arranged in order as a row vector. Subsequently, the distribution entropies of the IMF components of the same sample lithium battery are spliced at the head and tail according to the signal type, so as to obtain the corresponding feature vector of the sample lithium-ion battery.

[0100] Exemplarily, taking the voltage signal as an example, the row vector of the distribution entropy of the IMF components obtained by decomposing the voltage signal can be as shown in formula (8), and the feature vector of the sample lithium-ion battery obtained by aligning and splicing the head and tail according to the signal type can be as shown in formula (9).

[0101] DE v (IMF) = {DE v (IMF1), DE v (IMF2), …, DE v (IMFk)} (8)

[0102] DE(IMFS) = {DE v (IMF), DE t (IMF), DE f (IMF), DE r (IMF) (9)

[0103] Among them, DE v (IMF) is the row vector of the distribution entropy of the IMF component, DE(IMFS) is the feature vector of the sample lithium-ion battery, k is the number of MBKA-VMD decomposition modes, DE t (IMF) is the distribution entropy of the IMF component obtained by decomposing the temperature signal, DE f (IMF) is the distribution entropy of the IMF component obtained by decomposing the pressure signal, DE r (IMF) is the distribution entropy of the IMF component obtained by decomposing the internal resistance signal.

[0104] It should be noted that in the embodiments of the present application, the calculation methods of the temperature signal, the pressure signal, and the internal resistance signal are the same as those of the voltage signal.

[0105] In some embodiments, after the server determines the feature vector of the sample lithium-ion battery, the training set and the test set of the lithium-ion battery safety warning model can be determined according to the feature vector of the sample lithium-ion battery.

[0106] Exemplarily, the server can first generate a first candidate sample set according to the feature vector of the sample lithium-ion battery and the fault type label. The first candidate sample set includes sample data pairs composed of the feature vector and different fault type labels. Subsequently, the server can use the correlation coefficient method to perform feature screening on the sample data pairs in the first candidate sample set to obtain a second candidate sample set. The second candidate sample set includes target sample data pairs in which the screened feature vector is positively correlated with the fault type label. Finally, the server divides the training set and the test set from the second candidate sample set.

[0107] Among them, the fault type label can indicate the fault type of the sample lithium-ion battery or whether the sample lithium-ion battery is faulty. The embodiments of the present application do not limit this. The correlation coefficient method can be the Pearson correlation coefficient method.

[0108] Exemplarily, the server can assign corresponding fault type labels L to each feature vector, so that the sample data pairs composed of the feature vectors and different fault type labels are formed, thereby forming a first candidate sample set. Among them, the sample data pair can be as shown in formula (10).

[0109] Q = {DE(IMFS), L} (10)

[0110] Among them, Q is the sample data pair, DE(IMFS) is the feature vector, and L is the fault type label.

[0111] Exemplarily, Figure 3 is a schematic diagram of the distribution entropy value of the IMF component of the voltage signal provided by the embodiments of the present application. The distribution entropy values of the IMF components of the voltage signals of faulty batteries and normal batteries are different. Correspondingly, different fault type labels can be set for the feature vectors of the sample lithium-ion batteries based on the distribution entropy of the IMF components of the voltage signal.

[0112] Exemplarily, the server can use the Pearson correlation coefficient method to perform a correlation analysis on the sample data pairs in the first candidate sample set, and determine whether each sample data pair is a positive correlation sample pair, a negative correlation sample pair, or an uncorrelated sample pair. For example, if the Pearson correlation coefficient Ps of the sample data pair is > 0, the feature vector in the sample data pair is positively correlated with the fault type label, and the sample data pair is a positive correlation sample pair; if the Pearson correlation coefficient Ps of the sample data pair = 0, the feature vector in the sample data pair is uncorrelated with the fault type label, and the sample data pair is an uncorrelated sample pair; if the Pearson correlation coefficient Ps of the sample data pair is < 0, the feature vector in the sample data pair is negatively correlated with the fault type label, and the sample data pair is a negative correlation sample pair.

[0113] It should be understood that in the present application, after determining the correlation of the sample data pairs through the Pearson correlation coefficient method, the target sample data pairs in which the feature vectors are positively correlated with the fault type labels can be screened out. That is, the sample pairs with a correlation coefficient value greater than 0 are retained as the target sample data pairs, and the remaining sample data pairs are deleted, thereby forming a second candidate sample set.

[0114] Among them, the Pearson correlation coefficient method calculates the correlation of the sample data pairs, which can be realized by formula (11).

[0115]

[0116] Among them, b is the number of samples, DE i and Li is the i-th sample feature and the corresponding label, and is the mean of the i-th sample feature and label.

[0117] In some embodiments, the server can divide the second candidate sample set into a training set and a test set proportionally. The proportion can be specifically set according to the actual situation. For example, the training set is 60% and the test set is 40%; or, the training set is 70% and the test set is 30%.

[0118] In some embodiments, after determining the training set, the server can first use the improved black-winged kite optimization algorithm to optimize the hyperparameters of the random forest model to obtain the optimal parameters of the random forest model. Subsequently, the server can input the optimal parameters of the random forest model into the random forest model. Finally, the server can use the training set to train the random forest model with the input optimal parameters to obtain a lithium-ion battery safety warning model.

[0119] Exemplarily, Figure 4 is a schematic diagram of using the improved black-winged kite optimization algorithm to optimize the VMD algorithm provided by an embodiment of the present application. As Figure 4 shown, after the parameters of the improved black-winged kite optimization algorithm are initialized, the hyperparameters of the random forest model (RF) are optimized according to formulas (4) and (5) to obtain the optimal parameters. Subsequently, after the server inputs the optimal parameters into the random forest, it uses the training set data for training to obtain a safety warning model for lithium-ion batteries.

[0120] In some embodiments, after using the training set to complete the training of the safety warning model for lithium-ion batteries, the data in the test set can be input into the trained safety warning model for lithium-ion batteries to obtain a fault diagnosis result and perform a safety warning, thereby completing the test of the safety warning model for lithium-ion batteries.

[0121] In this application, the safety warning model for lithium-ion batteries considers the response of voltage to faults and combines multiple physical field signals such as temperature, pressure, and internal resistance to extract fault features from multivariate data, so as to more accurately diagnose battery faults and achieve safety warnings.

[0122] S203. Perform a safety warning on the lithium-ion battery according to the fault diagnosis result.

[0123] In this step, when the server determines the fault diagnosis result, it can perform a safety warning on the lithium-ion battery according to the fault diagnosis result.

[0124] In some embodiments, the fault diagnosis result can be used to determine the safety warning information of the lithium-ion battery.

[0125] Exemplarily, if the fault diagnosis result is that the lithium-ion battery has no fault, there is no need to issue a safety warning for the lithium-ion battery. Correspondingly, there is no need to generate a safety warning message for the lithium-ion battery. If the fault diagnosis result is that the lithium-ion battery has a fault, it is necessary to issue a safety warning for the lithium-ion battery. Correspondingly, the fault type label in the fault diagnosis result can be added to the safety warning message of the lithium-ion battery.

[0126] The safety warning method for a lithium-ion battery provided by an embodiment of the present application first obtains a plurality of signals to be measured during the operation of the lithium-ion battery; subsequently, inputs the plurality of signals to be measured into the lithium-ion battery safety warning model, and obtains the fault diagnosis result output by the lithium-ion battery safety warning model. The lithium-ion battery safety warning model is a model generated after being trained by a training set, and the training set is generated by processing the target multi-signal of the sample lithium-ion battery during operation using an improved black-winged kite optimization algorithm; finally, according to the fault diagnosis result, a safety warning is issued for the lithium-ion battery. Since the lithium-ion battery safety warning model can extract the characteristics of the multi-signal during the operation of the lithium-ion battery, the diversity of the characteristics is improved. At the same time, the improved black-winged kite optimization algorithm is used to process the data in the training set of the lithium-ion battery safety warning model, so that a more sufficient search can be performed around during the optimization process, avoiding falling into a local optimal solution, and thus the accuracy of the safety warning of the lithium-ion battery can be improved.

[0127] Figure 5 It is a schematic flowchart of a safety warning method for a lithium-ion battery provided by an embodiment of the present application, as Figure 5 shown. The safety warning method for the lithium-ion battery includes S301-S307:

[0128] S301. Collect the target multi-signal of the sample lithium-ion battery during operation.

[0129] In some embodiments, the server can collect the operation data of each battery cell of the sample lithium-ion battery during charge and discharge, and the sample lithium-ion battery is a battery pack formed by connecting a normal battery and a faulty battery in series. Subsequently, the server can form the operation data of each battery cell into a target multi-signal.

[0130] Among them, the above sample lithium-ion battery can electrically abuse a normal lithium-ion battery through overcharging, over-discharging, and parallel resistance between the positive and negative electrodes, etc., to make it a faulty battery, and then form a battery pack by connecting the normal battery and the faulty battery in series as the sample lithium-ion battery.

[0131] S302. Generate a training set and a test set for the lithium-ion battery safety warning model according to the target multi-signal.

[0132] In some embodiments, the server may first divide the target multivariate signal into multiple calculation windows. Secondly, the server uses an improved black-winged kite optimization algorithm to perform variational mode decomposition on the sub-signals in each calculation window to obtain multiple intrinsic mode function components of the target multivariate signal. Thirdly, the server calculates the distribution entropy of each intrinsic mode function component, and splices the distribution entropy of the intrinsic mode function components of the same sample lithium-ion battery at the head and tail according to the signal type to obtain the corresponding feature vector of the sample lithium-ion battery. Finally, the server determines the training set and the test set of the lithium-ion battery safety warning model according to the feature vector of the sample lithium-ion battery.

[0133] In some embodiments, after determining the training set, the server may first generate a first candidate sample set according to the feature vector of the sample lithium-ion battery and the fault type label. The first candidate sample set includes sample data pairs composed of feature vectors and different fault type labels. Subsequently, the server may use the correlation coefficient method to perform feature screening on the sample data pairs in the first candidate sample set to obtain a second candidate sample set. The second candidate sample set includes target sample data pairs in which the screened feature vectors are positively correlated with the fault type label. Finally, the server divides the training set and the test set from the second candidate sample set.

[0134] S303. Use the training set to train the lithium-ion battery safety warning model.

[0135] In some embodiments, after determining the training set, the server may first use an improved black-winged kite optimization algorithm to optimize the hyperparameters of the random forest model to obtain the optimal parameters of the random forest model. Subsequently, the server may input the optimal parameters of the random forest model into the random forest model. Finally, the server may use the training set to train the random forest model with the input optimal parameters to obtain the lithium-ion battery safety warning model.

[0136] S304. Use the test set to test the trained lithium-ion battery safety warning model.

[0137] S305. Obtain the multivariate signal to be measured during the operation of the lithium-ion battery.

[0138] S306. Input the multivariate signal to be measured into the lithium-ion battery safety warning model, and obtain the fault diagnosis result output by the lithium-ion battery safety warning model.

[0139] Among them, the lithium-ion battery safety warning model is a model generated after being trained by the training set. The training set is generated after processing the target multivariate signal of the sample lithium-ion battery during operation using an improved black-winged kite optimization algorithm.

[0140] S307. Perform safety warning on the lithium-ion battery according to the fault diagnosis result.

[0141] The safety warning method for a lithium-ion battery provided by an embodiment of the present application first obtains a plurality of signals to be measured during the operation of the lithium-ion battery; subsequently, inputs the plurality of signals to be measured into a lithium-ion battery safety warning model, and obtains a fault diagnosis result output by the lithium-ion battery safety warning model. The lithium-ion battery safety warning model is a model generated after being trained by a training set, and the training set is generated by processing the target plurality of signals of a sample lithium-ion battery during operation using an improved black kite optimization algorithm; finally, performs a safety warning on the lithium-ion battery according to the fault diagnosis result. Since the lithium-ion battery safety warning model can extract the characteristics of the plurality of signals during the operation of the lithium-ion battery, thereby improving the diversity of the characteristics, and at the same time uses the improved black kite optimization algorithm to process the data in the training set of the lithium-ion battery safety warning model, it can search more fully around during the optimization process, avoid falling into a local optimal solution, and thus can improve the accuracy of the safety warning of the lithium-ion battery.

[0142] It should be understood that although the steps in the flowcharts involved in the above embodiments are sequentially shown according to the indications of the arrows, these steps do not necessarily need to be executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps does not have a strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages. These steps or stages do not necessarily need to be executed at the same moment, but can be executed at different moments. The execution order of these steps or stages does not necessarily need to be sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0143] Based on the same inventive concept, an embodiment of the present application also provides a safety warning device for a lithium-ion battery for implementing the safety warning method for a lithium-ion battery involved above. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the safety warning device for a lithium-ion battery provided below can refer to the limitations on the safety warning method for a lithium-ion battery in the above text, and will not be repeated here.

[0144] In an exemplary embodiment, as Figure 6 shown, a safety warning device 400 for a lithium-ion battery is provided, including: an acquisition module 401, a diagnosis module 402, and a warning module 403, where:

[0145] The acquisition module 401 is configured to acquire a plurality of signals to be measured during the operation of the lithium-ion battery.

[0146] The diagnostic module 402 is configured to input the multi-variable signal to be measured into the lithium-ion battery safety warning model and obtain the fault diagnosis result output by the lithium-ion battery safety warning model. The lithium-ion battery safety warning model is a model generated after being trained by a training set, and the training set is generated by processing the target multi-variable signal of the sample lithium-ion battery during operation using the improved black-winged kite optimization algorithm.

[0147] The warning module 403 is configured to perform safety warning on the lithium-ion battery according to the fault diagnosis result.

[0148] In one embodiment, the safety warning device 400 for lithium-ion batteries further includes:

[0149] The training module 404 is configured to collect the target multi-variable signal of the sample lithium-ion battery during operation; generate a training set and a test set for the lithium-ion battery safety warning model according to the target multi-variable signal; use the training set to train the lithium-ion battery safety warning model; and use the test set to test the trained lithium-ion battery safety warning model.

[0150] In one embodiment, the training module 404 is further configured to divide the target multi-variable signal into multiple calculation windows; use the improved black-winged kite optimization algorithm to perform variational mode decomposition on the sub-signals in each calculation window to obtain multiple intrinsic mode function components of the target multi-variable signal; calculate the distribution entropy of each intrinsic mode function component; splice the distribution entropies of the intrinsic mode function components of the same sample lithium-ion battery at the head and tail according to the signal type to obtain the corresponding feature vector of the sample lithium-ion battery; and determine the training set and the test set of the lithium-ion battery safety warning model according to the feature vector of the sample lithium-ion battery.

[0151] In one embodiment, the training module is further configured to generate a first candidate sample set according to the feature vector of the sample lithium-ion battery and the fault type label. The first candidate sample set includes sample data pairs composed of the feature vector and different fault type labels; use the correlation coefficient method to perform feature screening on the sample data pairs in the first candidate sample set to obtain a second candidate sample set, and the second candidate sample set includes target sample data pairs in which the screened feature vector is positively correlated with the fault type label; and divide the training set and the test set from the second candidate sample set.

[0152] In one embodiment, the training module 404 is further configured to improve the initial black-winged kite optimization algorithm by fusing a dynamic reverse learning strategy and a golden sine strategy to obtain the improved black-winged kite optimization algorithm.

[0153] In one embodiment, the training module 404 is further configured to collect the operation data of each battery cell of the sample lithium-ion battery during operation in the charge and discharge state, where the sample lithium-ion battery is a battery pack formed by connecting a normal battery and a faulty battery in series; and form the operation data of each battery cell into a target multi-source signal.

[0154] In one embodiment, the diagnosis module 403 is further configured to optimize the hyperparameters of the random forest model using an improved black-winged kite optimization algorithm to obtain the optimal parameters of the random forest model; input the optimal parameters of the random forest model into the random forest model; and use the training set to train the random forest model with the input optimal parameters to obtain a lithium-ion battery safety warning model.

[0155] Each module in the above-mentioned lithium-ion battery safety warning device can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of the computer device in hardware form or be independent of it, or be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0156] In an exemplary embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 7 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a lithium-ion battery safety warning method.

[0157] Those skilled in the art can understand that Figure 7 the structure shown in

[0158] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the above-mentioned safety warning method for lithium-ion batteries is implemented.

[0159] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned safety warning method for lithium-ion batteries is implemented.

[0160] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the above-mentioned safety warning method for lithium-ion batteries is implemented.

[0161] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include Read-Only Memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, Resistive Random Access Memory (ReRAM), Magnetoresistive Random Access Memory (MRAM), Ferroelectric Random Access Memory (FRAM), Phase Change Memory (PCM), graphene memory, etc. Volatile memory can include Random Access Memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, Artificial Intelligence (AI) processors, etc., without limitation.

[0162] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered to be within the scope recorded in the present application.

[0163] The above embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A safety early warning method for lithium-ion batteries, characterized in that: The method comprises: Obtaining the multi-signals to be tested during the operation of lithium-ion batteries; Inputting the multivariate signal to be tested into a lithium-ion battery safety early warning model, and obtaining a fault diagnosis result output by the lithium-ion battery safety early warning model, wherein the lithium-ion battery safety early warning model is a model generated after training with a training set, wherein the training set is generated after processing target multivariate signals of a sample lithium-ion battery during operation using an improved black-winged kite optimization algorithm; According to the fault diagnosis result, a safety warning is given to the lithium-ion battery.

2. The method according to claim 1, characterized in that Before inputting the target multivariate signal into the lithium-ion battery safety early warning model and obtaining the fault diagnosis result output by the lithium-ion battery safety early warning model, the method further includes: Collecting target multivariate signals of the sample lithium-ion battery during operation; Generating a training set and a test set of the lithium-ion battery safety early warning model according to the target multivariate signal; Using the training set, training the lithium-ion battery safety early warning model; The trained lithium-ion battery safety warning model is tested using the test set.

3. The method according to claim 2, characterized in that The step of generating a training set and a test set of the lithium-ion battery safety early warning model according to the target multivariate signal comprises: Dividing the target multivariate signal into a plurality of calculation windows; Using the improved black-winged kite optimization algorithm, performing variational mode decomposition on the sub-signals in each calculation window to obtain multiple intrinsic mode function components of the target multivariate signal; Calculate the distribution entropy of each intrinsic mode function component; The distribution entropy of the intrinsic mode function components of the same sample lithium-ion battery is concatenated head to tail according to the signal type to obtain the characteristic vector of the corresponding sample lithium-ion battery; According to the characteristic vectors of the sample lithium-ion batteries, a training set and a test set of a lithium-ion battery safety early warning model are determined.

4. The method according to claim 3, characterized in that: The step of determining a training set and a test set of a lithium-ion battery safety warning model according to the characteristic vector of the sample lithium-ion battery comprises: Generate a first candidate sample set according to the feature vector and the fault type label of the sample lithium-ion battery, wherein the first candidate sample set includes sample data pairs consisting of the feature vector and different fault type labels; Using the correlation coefficient method to perform feature screening on the sample data pairs in the first candidate sample set to obtain a second candidate sample set, wherein the second candidate sample set includes target sample data pairs whose feature vectors are screened out and are positively correlated with the fault type label; The training set and the test set are divided from the second candidate sample set.

5. The method according to claim 3, characterized in that: The method further comprises: By integrating the dynamic reverse learning strategy and the golden sine strategy, the initial black kite optimization algorithm is improved to obtain the improved black kite optimization algorithm.

6. The method according to claim 2, characterized in that The method of collecting target multi-element signals of the lithium-ion battery during operation includes: Under the charge and discharge state, collecting the operation data of each cell of the sample lithium-ion battery during operation, wherein the sample lithium-ion battery is a battery pack formed by connecting a normal battery and a faulty battery in series; The operation data of each battery cell is used to form the target multi-element signal.

7. The method according to claim 2, characterized in that The using the training set to train the lithium-ion battery safety early warning model includes: Using the improved black kite optimization algorithm to optimize the hyperparameters of the random forest model to obtain the optimal parameters of the random forest model; Inputting the optimal parameters of the random forest model into the random forest model; The training set is used to train the random forest model with the optimal parameters input to obtain the lithium-ion battery safety warning model.

8. A safety warning device for lithium-ion batteries, characterized in that: The device comprises: An acquisition module is used to acquire the multi-signals to be tested during the operation of the lithium-ion battery; A diagnostic module, used for inputting the multivariate signal to be tested into a lithium-ion battery safety early warning model, and obtaining a fault diagnosis result output by the lithium-ion battery safety early warning model, wherein the lithium-ion battery safety early warning model is a model generated after training with a training set, wherein the training set is generated after processing target multivariate signals of a sample lithium-ion battery during operation using an improved black-winged kite optimization algorithm; The early warning module is used to provide a safety early warning for the lithium-ion battery according to the fault diagnosis result.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images