Battery thermal runaway dynamic prediction method, system, equipment, medium and product

Abstract

The invention discloses a battery thermal runaway dynamic prediction method, system and device, a medium and a product, and the method is characterized in that the method comprises the steps: obtaining a plurality of operation data of a lithium ion battery; inputting the operation data into a first auto-encoder for feature reconstruction to obtain first abnormal data, inputting the operation data into a second auto-encoder for feature reconstruction to obtain second abnormal data, and determining a first set, a second set and a target set based on the first abnormal data and the second abnormal data, respectively calculating relevancy between data features in the first set, the second set and the target set to obtain a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset, and determining target operation data based on the thermal runaway early warning feature subset, the thermal runaway sensitive feature subset and the target set; and performing thermal runaway prediction on the lithium ion battery based on the target operation data to obtain a thermal runaway prediction result of the lithium ion battery. According to the invention, the accuracy of battery thermal runaway prediction can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of battery management, and in particular to a battery thermal runaway dynamic prediction method, system, device, medium and product. BACKGROUND

[0002] With the wide application of lithium-ion batteries in electric vehicles, large-scale energy storage systems and portable electronic devices, the safety problems of batteries under high energy density and complex operating conditions have become increasingly prominent. Under the conditions of overcharge, overdischarge, external short circuit, internal defects or high temperature environment, severe exothermic reactions may occur inside the battery, which may further trigger thermal runaway, causing fire and even explosion and other serious safety accidents. Due to the characteristics of strong burst, fast evolution speed and large damage range of battery thermal runaway, if passive protection is only carried out after the accident occurs, it is often difficult to avoid safety risks in time. Therefore, how to dynamically and in advance predict and warn the potential thermal runaway risk during the operation of the battery has become a key technical problem that needs to be solved in the battery management system.

[0003] The existing battery thermal runaway prediction technology mainly includes a method based on a physical model and a data-driven method, but there are still obvious deficiencies in prediction accuracy. The method based on the physical model relies on the accurate modeling of the battery electrochemistry and thermal behavior, which is difficult to fully depict the nonlinear characteristics and individual differences of the battery under actual complex operating conditions. The model parameters are sensitive and the calculation complexity is high, which makes the prediction results easily affected by model errors. The data-driven method can learn features from historical data, but it usually needs a large number of real thermal runaway sample data covering different operating conditions and fault modes, and the actual thermal runaway sample acquisition cost is high and the number is limited, which leads to insufficient model generalization ability and easy misjudgment or omission. In addition, the existing methods mostly rely on a single anomaly discrimination mechanism, which is difficult to effectively distinguish between real anomalies and noise interference, thereby affecting the accurate identification and reliable prediction of the battery thermal runaway state. SUMMARY

[0004] The present application provides a battery thermal runaway dynamic prediction method, system, device, medium and product, which can improve the accuracy of battery thermal runaway prediction.

[0005] In a first aspect, the present application embodiment provides a battery thermal runaway dynamic prediction method, comprising:

[0006] Obtaining a plurality of operating data of a lithium-ion battery;

[0007] The operation data are input to a first autoencoder for feature reconstruction to learn the electro-thermal coupling change features deviating from normal working conditions in the operation data, and first abnormal data are obtained, the operation data are input to a second autoencoder for feature reconstruction to learn abnormal response features in the operation data, and second abnormal data are obtained, the correlation between the data features is calculated based on the feature distribution relationship between the first abnormal data and the second abnormal data, and the thermal runaway early warning feature subset and the thermal runaway sensitive feature subset are respectively screened based on the correlation, so that the target operation data for thermal runaway prediction are determined.

[0008] The lithium ion battery is predicted for thermal runaway based on the target operation data, and a thermal runaway prediction result of the lithium ion battery is obtained.

[0009] The embodiments of the present application comprehensively reflect the electrical state and thermal state changes of the battery in the actual operation process, provide multi-dimensional and continuous basic data support for subsequent thermal runaway feature mining, and improve the reliability of thermal runaway prediction; by starting from the normal operation benchmark, subtle coupling abnormal changes between parameters such as voltage, temperature and internal resistance in the early stage of thermal runaway are captured, so that potential thermal runaway precursors can be presented in advance, thereby improving the foresight and sensitivity of thermal runaway prediction; by modeling the sudden response and nonlinear amplification behavior in the battery operation data from the abnormal mode, the evolution law of typical abnormal features in the thermal runaway development process is effectively described, thereby enhancing the recognition ability of real abnormal behavior and reducing the probability of misjudgment; by distinguishing abnormal features appearing only in a single abnormal perspective from key abnormal features appearing in multiple abnormal representations, clear basis is provided for subsequent feature screening, thereby improving the accuracy of thermal runaway feature positioning; by quantitatively screening key features highly related to thermal runaway evolution from a large number of abnormal features, the interference of redundant information and noise features on the prediction model is reduced, thereby improving the effectiveness and discriminability of thermal runaway prediction features; by realizing the fusion expression of multi-source and multi-stage abnormal information, the data input to the prediction stage not only contains early warning information, but also reflects the abnormal evolution trend, thereby improving the integrity and accuracy of thermal runaway state representation as a whole, and significantly improving the accuracy and reliability of the thermal runaway prediction result of the lithium ion battery.

[0010] Further, the operation data are input to a first autoencoder for feature reconstruction to learn the electro-thermal coupling change features deviating from normal working conditions in the operation data, and first abnormal data are obtained, including:

[0011] The operation data are input to a first autoencoder for feature coding, and first reconstruction data corresponding to the operation data are obtained, wherein the first autoencoder is obtained based on the normal operation condition of the battery;

[0012] calculate a first reconstruction error between each of the operation data and each of the first reconstruction data;

[0013] compare the first reconstruction error with a first preset threshold, to filter each of the operation data based on a first comparison result, and determine the first abnormal data.

[0014] The embodiment of the application can effectively identify representative abnormal operation characteristics in the early stage of thermal runaway, reduce the interference of noise data and transient fluctuations, provide high-quality input data for subsequent thermal runaway feature fusion and prediction models, and thus improve the accuracy and reliability of battery thermal runaway prediction.

[0015] Further, the inputting each of the operation data into the second autoencoder for feature reconstruction to learn abnormal response features in each of the operation data to obtain second abnormal data comprises:

[0016] input each of the operation data into the second autoencoder for feature coding to obtain second reconstruction data corresponding to each of the operation data, wherein the second autoencoder is trained based on battery thermal runaway conditions;

[0017] calculate a second reconstruction error between each of the operation data and each of the second reconstruction data;

[0018] compare the second reconstruction error with a second preset threshold, to filter each of the operation data based on a second comparison result, and determine the second abnormal data.

[0019] The embodiment of the application can highlight abnormal response features highly related to thermal runaway evolution, and thus improve the accuracy and reliability of battery thermal runaway prediction.

[0020] Further, the calculating correlation between each data feature based on the feature distribution relationship between the first abnormal data and the second abnormal data comprises:

[0021] calculate the intersection of the first abnormal data and the second abnormal data to obtain the target set;

[0022] determine data in the first abnormal data except the target set as the first set, and determine data in the second abnormal data except the target set as the second set;

[0023] perform feature correlation degree calculation on data in the first set and data in the second set with data in the target set respectively to obtain first correlation results and second correlation results respectively.

[0024] The embodiment of the application performs intersection operation on the first abnormal data and the second abnormal data, so that the target set can represent key abnormal features highly related to battery thermal runaway; meanwhile, data identified as abnormal in a single model is divided into the first set and the second set respectively, reducing the risk of misjudgment caused by a single discrimination mechanism, providing a clear data structure basis for subsequent feature screening and fusion based on correlation, and further improving the accuracy and stability of battery thermal runaway prediction.

[0025] Further, the thermal runaway early warning feature subset and the thermal runaway sensitive feature subset are screened based on the correlation degrees respectively, including:

[0026] screening data in the first set according to the first correlation results to obtain a thermal runaway early warning feature subset;

[0027] screening data in the second set according to the second correlation results to obtain a thermal runaway sensitive feature subset.

[0028] The embodiment of the application calculates the correlation degrees between the data features in the first set and the second set and the key abnormal features in the target set respectively, so that data identified as abnormal in a single model can be re-screened according to its correlation degree with the core abnormal features of thermal runaway, providing more discriminative and time-consistent feature inputs for subsequent thermal runaway prediction models, and further improving the accuracy and reliability of battery thermal runaway prediction.

[0029] Further, the thermal runaway early warning feature subset and the thermal runaway sensitive feature subset are screened based on the correlation degrees respectively, including:

[0030] performing feature normalization processing on the target running data to obtain an abnormal feature sequence, wherein the target running data is all data in the thermal runaway early warning feature subset, the thermal runaway sensitive feature subset and the target set;

[0031] inputting the abnormal feature sequence into a preset thermal runaway prediction model for inference calculation to obtain a thermal runaway risk prediction value;

[0032] The thermal runaway risk prediction value is compared with a third preset threshold value to determine a thermal runaway risk level of the lithium ion battery based on a third comparison result.

[0033] The embodiment of the present application realizes quantitative prediction of the thermal runaway risk of the lithium ion battery by normalizing the target operation data and inputting the thermal runaway prediction model for inference calculation, and determines the corresponding risk level through threshold comparison, thereby improving the accuracy, stability and engineering practicability of the thermal runaway prediction result.

[0034] In a second aspect, the embodiment of the present application provides a battery thermal runaway dynamic prediction system, which comprises an acquisition module, a reconstruction module and a prediction module.

[0035] The acquisition module is configured to acquire a plurality of operation data of a lithium ion battery.

[0036] The reconstruction module is configured to input each of the operation data into a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, obtain first abnormal data, input each of the operation data into a second autoencoder for feature reconstruction to learn an abnormal response feature in each of the operation data, obtain second abnormal data, calculate a correlation degree between each data feature based on a feature distribution relationship between the first abnormal data and the second abnormal data, and filter a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation degree, so as to determine target operation data for thermal runaway prediction.

[0037] The prediction module is configured to predict thermal runaway of the lithium ion battery based on the target operation data, and obtain a thermal runaway prediction result of the lithium ion battery.

[0038] The embodiment of the present application sets the acquisition module, the reconstruction module and the prediction module, uses the first autoencoder and the second autoencoder to respectively reconstruct features of the battery operation data from two angles of the normal working condition and the thermal runaway working condition, and analyzes and filters key features with early warning significance and sensitivity in combination with the feature distribution relationship and the correlation degree, so as to construct the target operation data for thermal runaway prediction, realize accurate and stable prediction of the thermal runaway risk of the lithium ion battery, and improve the prediction reliability and engineering application value of the system under complex working conditions.

[0039] In a third aspect, the embodiment of the present application provides a terminal device, which comprises a processor, a memory, a communication interface and a communication bus, and the processor, the memory and the communication interface complete communication with each other through the communication bus.

[0040] The memory is configured to store at least one executable instruction, and the executable instruction is configured to enable the processor to perform the operation of the battery thermal runaway dynamic prediction method.

[0041] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which comprises a stored computer program, wherein the computer program controls a device or a system where the computer readable storage medium is located to perform the battery thermal runaway dynamic prediction method when the computer program is executed.

[0042] On the basis of the above-mentioned method embodiment, another embodiment of the present application provides a computer program product, which comprises a computer program or instructions, and the computer program or instructions are executed by a communication device to realize the battery thermal runaway dynamic prediction method of any one of the embodiments of the present application.

[0043] The above description is only a summary of the technical solutions of the embodiments of the present application, in order to more clearly understand the technical means of the embodiments of the present application, the embodiments of the present application can be implemented according to the content of the description, and in order to make the above and other purposes, characteristics and advantages of the embodiments of the present application more obvious and easy to understand, the specific embodiments of the present application are described below. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the present application, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 is a flowchart of an embodiment of the battery thermal runaway dynamic prediction method provided by the present application;

[0046] Figure 2 is a flowchart of steps S201 to S203 provided by the present application;

[0047] Figure 3 is a flowchart of steps S301 to S303 provided by the present application;

[0048] Figure 4 is a flowchart of steps S401 to S402 provided by the present application;

[0049] Figure 5 is a structural schematic diagram of an embodiment of the battery thermal runaway dynamic prediction method provided by the present application. DETAILED DESCRIPTION

[0050] In order to make the objects, technical solutions and advantages of the present application clearer, the following will clearly and completely describe the technical solutions in the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort should fall into the scope of the present application.

[0051] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of this application; the use of the terms "including," "comprising," "having" and "with" and any variations thereof in this specification and in the claims are intended to cover both the inclusive and exclusive cases.

[0052] In the description of the embodiments of the present application, the technical terms "first", "second" and the like are only used to distinguish different objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number, specific order or primary and secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0053] Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearance of the phrase in various places in the specification are not necessarily all referring to the same embodiment, or are necessarily referring to different or alternative embodiments. It will be explicitly understood by a person of ordinary skill in the art that the embodiments described herein can be combined with other embodiments.

[0054] In the description of the embodiments of the present application, the term "and / or" is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent the three cases of A alone, A and B together, and B alone. In addition, the character " / " in the present application generally represents an "or" relationship between the associated objects.

[0055] In the description of the embodiments of the present application, the term "a plurality of" refers to two or more (including two), and similarly, "a plurality of groups" refers to two or more groups (including two groups), and "a plurality of pieces" refers to two or more pieces (including two pieces).

[0056] In the description of the embodiments of the present application, unless otherwise explicitly specified and limited, the technical terms "mounting", "connection", "connecting", "fixing" and the like should be understood in a broad sense, for example, can be fixed connection, can also be detachable connection, or integral; can be mechanical connection, can also be electrical connection; can be direct connection, can also be indirect connection through an intermediate medium, can be internal communication of two elements or interaction relationship between two elements. For those skilled in the art, the specific meanings of the above terms in the embodiments of the present application can be understood according to the specific circumstances.

[0057] With the wide application of lithium-ion batteries in electric vehicles, large-scale energy storage systems and portable electronic devices, the safety risks under high energy density and complex operating conditions are increasingly prominent. Under the conditions of overcharge, overdischarge, short circuit, internal defects or high temperature environment, severe exothermic reactions may occur inside the battery and cause thermal runaway, resulting in fire or even explosion and other serious accidents. Due to the characteristics of strong burst and fast evolution of battery thermal runaway, it is difficult to avoid risks in time by relying on post-protection only, so it is urgent to realize dynamic prediction and early warning of thermal runaway risk during operation. The existing thermal runaway prediction methods mainly include physical model-based and data-driven methods, wherein the physical model method is difficult to accurately describe the nonlinear characteristics under complex conditions, and has high computational complexity; the data-driven method is limited by the scarcity of thermal runaway samples and single abnormality discrimination mechanism, and is prone to misjudgment or omission, resulting in insufficient prediction accuracy and reliability.

[0058] Referring to Figure 1 To improve the accuracy of battery thermal runaway prediction, an embodiment of the present application provides a battery thermal runaway dynamic prediction method, comprising steps S101 to S103.

[0059] Step S101, obtaining a plurality of operating data of a lithium-ion battery;

[0060] In some embodiments, the method comprises obtaining a plurality of operating data of the lithium ion battery, including collecting multi-dimensional state parameters of the target battery during operation in real time through a sensor module, a collection module and a communication module in a battery management system (BMS) to form an original operating data set for subsequent anomaly identification and thermal runaway prediction. The operating data at least includes one or more of electrical parameters, thermal parameters and characteristic parameters reflecting the internal state change of the battery, wherein the electrical parameters include the terminal voltage, current, internal resistance, voltage change rate, current fluctuation amplitude, etc. of the battery monomer or battery module; the thermal parameters include the battery surface temperature, internal temperature estimation value, temperature gradient, temperature change rate, etc.; the characteristic parameters reflecting the internal state change of the battery include the state of charge (SOC), the state of health (SOH), the cell expansion amount, the strain change amount, the gas generation characteristic parameters and the gas concentration data of CO, CO2, H2, etc. obtained based on the gas sensor. The operating data is continuously collected according to a preset sampling period, and the sampling period can be 10 ms, 100 ms, 1 s or adaptively set according to the battery type and application scenario to ensure that the operating data can completely reflect the dynamic change process of the battery under different working conditions.

[0061] In some embodiments, after obtaining the operating data, the method further comprises a step of preprocessing the operating data, including but not limited to data time alignment, abnormal missing value compensation, noise filtering and normalization processing. The normalization processing is used to map the operating data of different dimensions to a unified numerical interval to reduce the influence of different parameter scale differences on the subsequent feature learning process of the autoencoder. The operating data can be organized according to a time window to construct a time series data segment containing a plurality of continuous sampling points as the basic data unit for subsequent input to the normal autoencoder and the abnormal autoencoder, so that the model can learn the time evolution characteristics of the battery operating state.

[0062] Through the above steps, the multi-dimensional operating data of the lithium ion battery can be comprehensively and continuously obtained, and the noise is eliminated, the missing values are compensated and the different parameter scales are normalized through preprocessing, to construct high-quality time series data segments, provide a reliable data basis for the feature learning of the subsequent autoencoder, and thus improve the accuracy and real-time performance of the battery anomaly identification and thermal runaway prediction.

[0063] Step S102, input each of the operation data to a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, obtain first abnormal data, input each of the operation data to a second autoencoder for feature reconstruction to learn an abnormal response feature in each of the operation data, obtain second abnormal data, calculate a correlation between each data feature based on a feature distribution relationship of the first abnormal data and the second abnormal data, and filter a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation to determine target operation data for thermal runaway prediction.

[0064] Please refer to Figure 2 In some embodiments, the input of each of the operation data to the first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data to obtain first abnormal data comprises steps S201 to S203.

[0065] Step S201, input each of the operation data to a first autoencoder for feature coding to obtain first reconstruction data corresponding to each of the operation data, wherein the first autoencoder is trained based on a normal battery operation condition.

[0066] In some embodiments, multi-source real-time operation data of a target battery during operation are input to a first autoencoder for feature coding and decoding processing to obtain first reconstruction data corresponding to each of the operation data. The operation data can include voltage, current, surface temperature, internal resistance, SOC, temperature change rate, temperature difference, and gas production characteristics, and other parameter data reflecting the electro-thermal coupling state of the battery. The first autoencoder is a normal autoencoder, which is trained only with historical operation data collected when the battery is in a normal operation condition in the training stage, and the reconstruction error between the input data and the reconstruction data is minimized as the training target, so that the first autoencoder can learn and depict the electro-thermal coupling feature distribution law under the normal operation state of the battery.

[0067] In some embodiments, the first autoencoder includes an encoder and a decoder. The encoder is used to map high-dimensional operation data to a low-dimensional latent feature space to extract core features representing the normal working condition of the battery. The decoder is used to reconstruct the input operation data according to the latent features and output first reconstruction data consistent with the dimension of the input operation data. When the real-time operation data conforms to the normal condition distribution, the first autoencoder can reconstruct it with high precision. When the operation data deviates from the normal condition, its reconstruction ability will decrease significantly. The specific hierarchical structure and parameter configuration of the first autoencoder are shown in Table 1.

[0068] Table 1 Specific hierarchical structure and parameter configuration of the first autoencoder

[0069]

[0070] Step S202, calculating a first reconstruction error between each of the operation data and each of the first reconstruction data;

[0071] In some embodiments, the first reconstruction error can be obtained by calculating the difference between each of the operation data and the corresponding first reconstruction data, and can be calculated in the form of Euclidean distance, absolute error, mean square error or weighted error, and the calculation method is as follows:

[0072]

[0073] wherein, X i represents the i-th real-time operation data, represents the first reconstruction data output by the first autoencoder for the i-th operation data, E i represents the corresponding first reconstruction error; ‖·‖ represents a norm, which can be a Euclidean norm, a weighted Euclidean norm or an absolute error and form, and can be set according to the type of operation data. In this embodiment, the first reconstruction error of each operation data can be calculated respectively, so as to obtain a first reconstruction error sequence corresponding to each operation data. The numerical value of the first reconstruction error is used to represent the deviation degree of the corresponding operation data relative to the normal operation distribution, and the larger the first reconstruction error is, the more likely the operation data does not conform to the normal operation state characteristics.

[0074] Step S203, comparing the first reconstruction error with a first preset threshold, and screening each of the operation data based on a first comparison result to determine the first abnormal data.

[0075] In some embodiments, each of the first reconstruction errors obtained in step S202 is compared with a first preset threshold one by one, and each real-time operation data of the target battery is screened according to the comparison result to determine the first abnormal data. In this embodiment, the first preset threshold can be pre-set according to the reconstruction error statistical characteristics under the normal operation condition of the battery, for example, determined based on the mean and standard deviation of the normal sample reconstruction error, or obtained by adaptive updating of the historical operation data. Specifically, if the first reconstruction error corresponding to a real-time operation data is greater than the first preset threshold, it indicates that the operation data deviates from the normal operation characteristic distribution learned by the first autoencoder, and the real-time operation data is determined as the first abnormal data; if the first reconstruction error corresponding to a real-time operation data is less than or equal to the first preset threshold, it is considered that the operation data is still within the normal working condition range, and is not the first abnormal data.

[0076] Please refer to Figure 3In some embodiments, the inputting each of the operation data into the second autoencoder for feature reconstruction to learn an abnormal response feature in each of the operation data to obtain second abnormal data comprises steps S301-S303.

[0077] In step S301, each of the operation data is input into the second autoencoder for feature encoding to obtain second reconstruction data corresponding to each of the operation data, wherein the second autoencoder is trained based on a battery thermal runaway working condition.

[0078] In some embodiments, each of the real-time operation data of the target battery during operation is input into the second autoencoder for feature encoding and decoding processing to obtain second reconstruction data corresponding to each of the operation data. The operation data can include but is not limited to voltage, current, surface temperature, temperature change rate, internal resistance, SOC, pressure difference, temperature difference, gas concentration change amount, and derived feature data reflecting the degree of abnormal reaction of the battery. The second autoencoder is an abnormal autoencoder, which is trained in the training stage using operation data in the thermal runaway state fitted based on a battery physical model, and takes minimizing the reconstruction error between the input data and the reconstruction data as the training target, so that the second autoencoder can learn and represent the typical operation feature distribution of the battery in the thermal runaway working condition.

[0079] In some embodiments, the second autoencoder includes an encoder and a decoder. The encoder is used to map high-dimensional operation data to a low-dimensional latent feature space to extract core abnormal features representing the thermal runaway state of the battery. The decoder is used to reconstruct the input operation data according to the latent features to output second reconstruction data consistent with the dimension of the input operation data. Since the second autoencoder has been trained for the thermal runaway working condition, when the real-time operation data matches the thermal runaway state feature distribution, its reconstruction ability is strong. When the real-time operation data does not belong to the thermal runaway evolution feature, its reconstruction error will significantly increase. The specific hierarchical structure and parameter configuration of the second autoencoder are shown in Table 2.

[0080] Table 2 Specific hierarchical structure and parameter configuration of the second autoencoder

[0081]

[0082] In step S302, a second reconstruction error between each of the operation data and each of the second reconstruction data is calculated.

[0083] In some embodiments, the second reconstruction error corresponding to each of the operation data is calculated by performing a difference operation between each of the operation data obtained in step S301 and the corresponding second reconstruction data, so as to quantify the matching degree between the real-time operation data and the thermal runaway working condition feature distribution. The second reconstruction error can be calculated in the following manner:

[0084]

[0085] wherein X i represents the ith real-time operation data, represents the second reconstruction data output by the second autoencoder for the ith operation data, represents the corresponding second reconstruction error, and ‖·‖ represents a norm, which can be an Euclidean norm, a weighted Euclidean norm or an absolute error sum form; in the present embodiment, weights can be set for each dimension error according to the importance of different operation parameters, so as to improve the sensitivity to abnormality of temperature, voltage or gas production key parameters.

[0086] In step S303, the second reconstruction error is compared with a second preset threshold, so as to filter each of the operation data based on a second comparison result, and determine the second abnormal data.

[0087] In some embodiments, each of the second reconstruction errors obtained in step S302 is compared with a second preset threshold, and each of the real-time operation data of the target battery is filtered according to the comparison result, so as to determine the second abnormal data. In the present embodiment, the second preset threshold can be preset according to the statistical characteristics of the reconstruction error of the second autoencoder on the thermal runaway training sample, for example, determined based on the mean, variance or quantile of the reconstruction error, or dynamically adjusted in combination with the historical operation data. Specifically, when the second reconstruction error corresponding to a certain real-time operation data is less than or equal to the second preset threshold, it indicates that the operation data is highly matched with the thermal runaway working condition feature distribution learned by the second autoencoder, and thus the real-time operation data is determined as a second abnormal data; when the second reconstruction error corresponding to a certain real-time operation data is greater than the second preset threshold, it indicates that the operation data does not conform to the thermal runaway working condition feature distribution, and thus the operation data is excluded from the second abnormal data.

[0088] In some embodiments, the correlation between each data feature is calculated based on the feature distribution relationship between the first abnormal data and the second abnormal data, including: calculating the intersection of the first abnormal data and the second abnormal data to obtain the target set; determining the data in the first abnormal data except the target set as the first set, and determining the data in the second abnormal data except the target set as the second set; and performing feature correlation calculation on the data in the first set and the data in the second set with the data in the target set respectively to obtain first and second correlation results respectively.

[0089] In some embodiments, the intersection of the first abnormal data and the second abnormal data is calculated to obtain the target set, specifically: assuming that the real-time running data of the target battery in a certain time period includes multiple feature data such as voltage, current, temperature, internal resistance and gas production. After the above steps, the first abnormal data set A is obtained as A = {V1, T2, I3, R4, G5}, and the second abnormal data set B is obtained as B = {T2, I3, T6, G7}; the intersection of the first abnormal data set A and the second abnormal data set B is calculated: intersection C = A∩B = {T2, I3}; wherein T2 represents temperature abnormal data No. 2, and I3 represents current abnormal data No. 3. The intersection C is taken as the target set, i.e. the target set = {T2, I3}, which is used for subsequent thermal runaway prediction data processing.

[0090] In some embodiments, the data in the first abnormal data except the target set is determined as the first set, and the data in the second abnormal data except the target set is determined as the second set, specifically: according to the above example, the first abnormal data set A = {V1, T2, I3, R4, G5}, the target set C = {T2, I3}, and the data in A except the target set is determined as the first set D1, i.e. D1 = {V1, R4, G5}; the second abnormal data set B = {T2, I3, T6, G7}, the target set C = {T2, I3}, and the data in B except the target set is determined as the second set D2, i.e. D2 = {T6, G7}. In subsequent implementation, correlation calculation and screening can be performed on each data in the first set D1 and the second set D2 with the target set C respectively, and the candidate data screened is merged with the target set C to obtain the final target abnormal data set for thermal runaway prediction.

[0091] In some embodiments, the data in the first set and the data in the second set are respectively correlated with the data in the target set to obtain first correlation results and second correlation results, specifically: first, a correlation calculation model based on feature dimensions and time dimensions is constructed according to the first set and the target set obtained after the above steps, and feature correlation analysis is performed on the abnormal operation data in the first set and the abnormal operation data in the target set to quantify the correlation degree between the data in the first set and the confirmed abnormal mode. Specifically, the first set is represented as The target set is represented as Wherein, x is a multi-dimensional feature vector composed of voltage, current, temperature, temperature rate of change, internal resistance, gas concentration and the like. In this embodiment, the first correlation result is calculated by using the Pearson correlation coefficient, and the calculation formula is as follows:

[0092]

[0093] Wherein, r ij represents the first correlation result between the i-th data in the first set and the j-th data in the target set, and K is the number of feature dimensions. Secondly, according to the second set and the target set obtained after the above steps, further feature correlation calculation is performed to evaluate the matching degree between the data in the second set and the typical thermal runaway abnormal mode. The second set is represented as In this embodiment, the second correlation result is also calculated based on the Pearson correlation coefficient, and the calculation formula is consistent with the calculation of the Pearson correlation coefficient, only replacing the first set with the second set. Since the second set is derived from the abnormal autoencoder, the data is closer to the thermal runaway state in terms of distribution form, so the second correlation result is usually used to characterize the abnormal severity and abnormal evolution consistency. In this embodiment, a weight coefficient can also be introduced in the correlation calculation, and higher weight is given to features such as temperature and gas concentration which are more sensitive to thermal runaway, so as to improve the representation ability of the second correlation result to key runaway features.

[0094] Please refer to Figure 4 In some embodiments, the correlation between the features of the data in the first set, the second set and the target set is calculated respectively to obtain a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation, including steps S401 to S402.

[0095] Step S401, filtering the data in the first set according to the first correlation result to obtain a thermal runaway early warning feature subset;

[0096] In some embodiments, based on the first correlation results obtained in the foregoing steps, the abnormal operation data in the first set is further screened to eliminate noise data with low correlation with thermal runaway abnormalities. In this embodiment, a first correlation threshold T1 is preset. When the first correlation result corresponding to at least one data in the target set meets r ij ≥ T1, it is considered that the data has strong correlation with the thermal runaway abnormal characteristics, and is retained. After screening, a thermal runaway early warning characteristic subset is obtained, denoted as The thermal runaway early warning characteristic subset mainly includes data in the early stage of thermal runaway evolution, which has not yet fully met the thermal runaway abnormal distribution, but has been highly correlated with the target set in terms of characteristic change trend, and is used to capture potential risk signals in advance.

[0097] Step S402, screening the data in the second set according to the second correlation result to obtain a thermal runaway sensitive characteristic subset.

[0098] In some embodiments, based on the second correlation results obtained in the foregoing steps, the abnormal operation data in the second set is screened to extract abnormal characteristic data most sensitive to thermal runaway. In this embodiment, a second correlation threshold T2 is set. When the second correlation result corresponding to at least one data in the target set meets r ij ≥ T2, the data is determined as highly thermal runaway related data and is included in the thermal runaway sensitive characteristic subset. The final thermal runaway sensitive characteristic subset is denoted as The thermal runaway sensitive characteristic subset mainly reflects the key abnormal characteristics of the battery that has entered the rapid evolution stage of thermal runaway, and can be used as core input data for subsequent thermal runaway prediction and risk classification together with the thermal runaway early warning characteristic subset and the target set.

[0099] Through the above steps, the early abnormal characteristics of the battery deviating from the normal working condition and the typical abnormal characteristics of thermal runaway can be extracted by using the normal autoencoder and the abnormal autoencoder respectively, and the thermal runaway early warning characteristic subset and the thermal runaway sensitive characteristic subset can be effectively screened through the intersection of the sets and the correlation degree analysis, so as to realize the hierarchical characterization of the abnormal state of the battery and provide core characteristic data with high correlation and high sensitivity for subsequent accurate prediction of thermal runaway risk.

[0100] Step S103, performing thermal runaway prediction on the lithium ion battery based on the target operation data to obtain a thermal runaway prediction result of the lithium ion battery.

[0101] In some embodiments, the thermal runaway prediction of the lithium ion battery based on the target operation data comprises: performing feature normalization processing on the target operation data to obtain an abnormal feature sequence; inputting the abnormal feature sequence into a preset thermal runaway prediction model for inference calculation to obtain a thermal runaway risk prediction value; and comparing the thermal runaway risk prediction value with a third preset threshold to determine a thermal runaway risk level of the lithium ion battery based on a third comparison result.

[0102] In some embodiments, the target operation data is the abnormal feature sequence obtained by performing feature normalization processing on the thermal runaway early warning feature subset, the thermal runaway sensitive feature subset, and all data in the target set. Specifically, before performing thermal runaway prediction based on the target abnormal operation data, the target abnormal operation data is first processed by feature normalization to eliminate the influence of differences in dimensions, value ranges, and variation amplitudes of different operation parameters on the inference results of subsequent prediction models, thereby improving the prediction stability and accuracy. Specifically, the target abnormal operation data can be represented as: X = {x1, x2, …, xN}, where xN= [x1, x2, …, xM] is a multi-dimensional abnormal operation feature vector corresponding to the t-th sampling time point, M represents the number of abnormal feature dimensions, and the abnormal features include but are not limited to abnormal voltage features, abnormal current features, abnormal temperature features, temperature change rate features, internal resistance change features, and gas production change features. T t t,1 t,2 t,M In this embodiment, the target abnormal operation data is processed by the minimum-maximum normalization method, and the normalization formula is as follows:

[0103]

[0104]

[0105] ​​​​​​​​​In some embodiments, the abnormal feature sequence is input into a preset thermal runaway prediction model for inference calculation to obtain a thermal runaway risk prediction value. Specifically, after the construction of the abnormal feature sequence is completed, the abnormal feature sequence is input into a preset thermal runaway prediction model for inference calculation to obtain a risk prediction value of the target battery for thermal runaway in a future time window. In this embodiment, the thermal runaway prediction model is a time series prediction model based on deep learning, and the model includes a time feature extraction network and a risk mapping layer. The time feature extraction network is used to learn the time series dependence and evolution law in the abnormal feature sequence, and the risk mapping layer is used to map the extracted high-dimensional time series features to the corresponding thermal runaway risk prediction value. For example, the thermal runaway prediction model can be a long short-term memory network (LSTM), a gated recurrent unit (GRU), a time convolution network (TCN), or a combination structure of the above models, but the present application is not limited thereto. When the abnormal feature sequence is input into the thermal runaway prediction model, the model obtains the corresponding thermal runaway risk prediction value R through forward propagation calculation, and the representation form is as follows: wherein f(·) represents a nonlinear mapping function of the thermal runaway prediction model, and the thermal runaway risk prediction value R is a continuous numerical value, which is used to represent the risk degree of the target battery for thermal runaway in the prediction time range. In this embodiment, the thermal runaway risk prediction value can be represented as a thermal runaway occurrence probability, and the value range is [0, 1].

[0106] In some embodiments, the thermal runaway risk prediction value is compared with a third preset threshold to determine the thermal runaway risk level of the lithium ion battery based on a third comparison result. Specifically, after obtaining the thermal runaway risk prediction value, the thermal runaway risk prediction value is compared with a third preset threshold to determine the thermal runaway risk level of the target battery. In this embodiment, the third preset threshold includes at least two thresholds, i.e., a low risk threshold T L and a high risk threshold T H , and satisfies: 0≤T L <T H ≤1. According to the comparison result, the thermal runaway risk level of the target battery is determined, specifically including: when R<T L , it is determined that the target battery is in a low risk state; when T L ≤R<T H , it is determined that the target battery is in a medium risk warning state; and when R≥T H , it is determined that the target battery is in a high risk thermal runaway state. Through the above method, the continuous thermal runaway risk prediction value can be converted into an explicit risk level result, thereby providing intuitive and executable thermal runaway warning information for the battery management system.

[0107] ​Through the above steps, the abnormal feature sequence of a unified scale can be constructed based on the target abnormal operation data, the future thermal runaway risk of the battery is quantitatively predicted by using the preset deep learning thermal runaway prediction model, and the risk level is divided clearly by combining the threshold determination, so that the thermal runaway risk of the battery can be evaluated in real time and accurately, and a reliable basis is provided for safety management and early warning.

[0108] As Figure 5 indicated, on the basis of the above method embodiment, corresponding device embodiments are provided.

[0109] An embodiment of the present application provides a structural schematic diagram of a battery thermal runaway dynamic prediction system, which comprises an acquisition module 100, a reconstruction module 200 and a prediction module 300.

[0110] The acquisition module 100 is used for acquiring a plurality of operation data of a lithium ion battery.

[0111] The reconstruction module 200 is used for inputting each operation data into a first autoencoder for feature reconstruction, so as to learn the electro-thermal coupling change feature deviating from the normal working condition in each operation data, obtain first abnormal data, input each operation data into a second autoencoder for feature reconstruction, so as to learn the abnormal response feature in each operation data, obtain second abnormal data, calculate the correlation degree between each data feature based on the feature distribution relationship of the first abnormal data and the second abnormal data, and filter a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation degree respectively, so as to determine target operation data for thermal runaway prediction.

[0112] The prediction module 300 is used for predicting thermal runaway of the lithium ion battery based on the target operation data, and obtaining a thermal runaway prediction result of the lithium ion battery.

[0113] It can be understood that the above device embodiment corresponds to the method embodiment of the present application, and can realize the battery thermal runaway dynamic prediction method provided by any one of the above method embodiments. The more detailed working process and principle of the system can be but not limited to the related description of the above method.

[0114] It should be noted that the device embodiments described above are only schematic, and part or all of the modules can be selected to achieve the purpose of the embodiment. In addition, in the device embodiment provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which can be realized as one or more communication buses or signal lines. Those skilled in the art can understand and implement it without creative labor.

[0115] On the basis of the above-mentioned battery thermal runaway dynamic prediction method, another embodiment of the present application provides a terminal device, which comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, and when the computer program is executed by the processor, the battery thermal runaway dynamic prediction method of any one of the embodiments of the present application is realized.

[0116] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the terminal device.

[0117] The terminal device can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The terminal device can include, but is not limited to, a processor and a memory.

[0118] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, and connects all parts of the terminal device through various interfaces and lines.

[0119] On the basis of the above-mentioned method embodiment, another embodiment of the present application provides a computer readable storage medium, which comprises a stored computer program, wherein when the computer program runs, the device where the computer readable storage medium is located executes the battery thermal runaway dynamic prediction method of any one of the above-mentioned method embodiments of the present application.

[0120] The modules / units integrated in the device / terminal equipment, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0121] The above specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only for specific embodiments of the present application and is not intended to limit the protection scope of the present application. It is particularly pointed out that any modification, equivalent replacement, improvement, etc. made by those skilled in the art within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A battery thermal runaway dynamic prediction method, characterized in that, The method comprises: acquiring a plurality of operation data of a lithium ion battery; inputting each of the operation data into a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, and obtaining first abnormal data, inputting each of the operation data into a second autoencoder for feature reconstruction to learn an abnormal response feature in each of the operation data, and obtaining second abnormal data, calculating a correlation degree between each data feature based on a feature distribution relationship between the first abnormal data and the second abnormal data, and screening a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation degree, so as to determine target operation data for thermal runaway prediction; performing thermal runaway prediction on the lithium ion battery based on the target operation data, and obtaining a thermal runaway prediction result of the lithium ion battery.

2. The battery thermal runaway dynamic prediction method of claim 1, wherein, The method comprises: inputting each of the operation data into a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, and obtaining first abnormal data, inputting each of the operation data into a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, and obtaining first abnormal data, inputting each of the operation data into a second autoencoder for feature reconstruction to learn an abnormal response feature in each of the operation data, and obtaining second abnormal data, calculating a correlation degree between each data feature based on a feature distribution relationship between the first abnormal data and the second abnormal data, and screening a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation degree, so as to determine target operation data for thermal runaway prediction; The method comprises: inputting each of the operation data into a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, and obtaining first abnormal data, inputting each of the operation data into a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, and obtaining first abnormal data, inputting each of the operation data into a second autoencoder for feature reconstruction to learn an abnormal response feature in each of the operation data, and obtaining second abnormal data, calculating a correlation degree between each data feature based on a feature distribution relationship between the first abnormal data and the second abnormal data, and screening a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation degree, so as to determine target operation data for thermal runaway prediction; 3. The battery thermal runaway dynamic prediction method of claim 1, wherein, The method comprises: calculating an intersection of the first abnormal data and the second abnormal data to obtain the target set; determining data in the first set except the target set as the first set, and determining data in the second set except the target set as the second set; performing feature correlation degree calculation on data in the first set and data in the second set with data in the target set respectively to obtain first correlation results and second correlation results respectively.

4. The battery thermal runaway dynamic prediction method of claim 1, wherein, The method comprises: screening data in the first set according to the first correlation results to obtain a thermal runaway early warning feature subset; ​ ​ 5. The battery thermal runaway dynamic prediction method of claim 1, wherein, ​ ​ Screening data in the second set according to the second correlation result to obtain a thermal runaway sensitive feature subset.

6. The battery thermal runaway dynamic prediction method of claim 1, wherein, The thermal runaway prediction of the lithium ion battery based on the target operation data obtains a thermal runaway prediction result of the lithium ion battery, including: Feature normalization processing is performed on the target operation data to obtain an abnormal feature sequence, wherein the target operation data is all data in the thermal runaway early warning feature subset, the thermal runaway sensitive feature subset and the target set; The abnormal feature sequence is input into a preset thermal runaway prediction model for inference calculation to obtain a thermal runaway risk prediction value; The thermal runaway risk prediction value is compared with a third preset threshold to determine the thermal runaway risk level of the lithium ion battery based on a third comparison result.

7. A battery thermal runaway dynamic prediction system, comprising: The system comprises an acquisition module, a reconstruction module and a prediction module; The acquisition module is configured to acquire a plurality of operation data of a lithium ion battery. The reconstruction module is configured to input each of the operation data into a first autoencoder for feature reconstruction to learn an electro-thermal coupling change feature deviating from a normal working condition in each of the operation data, to obtain first abnormal data, input each of the operation data into a second autoencoder for feature reconstruction to learn an abnormal response feature in each of the operation data, to obtain second abnormal data, and based on a feature distribution relationship between the first abnormal data and the second abnormal data, to calculate a correlation degree between each data feature, to screen a thermal runaway early warning feature subset and a thermal runaway sensitive feature subset based on the correlation degree, and to determine target operation data for thermal runaway prediction. The prediction module is configured to predict the thermal runaway of the lithium ion battery based on the target operation data to obtain a thermal runaway prediction result of the lithium ion battery.

8. A terminal device, comprising: A computer program is stored in the memory and configured to be executed by the processor, and when the computer program is executed by the processor, the battery thermal runaway dynamic prediction method of any one of claims 1-6 is implemented.

9. A computer-readable storage medium, characterized in that, It includes: The computer program or instruction is executed by the communication device to implement the battery thermal runaway dynamic prediction method of any one of claims 1 to 6.

10. A computer program product comprising computer programs or instructions, characterized in that, The computer program or instruction is executed by the communication device to implement the battery thermal runaway dynamic prediction method of any one of claims 1 to 6.

Images