CN-122017603-A - Battery fault early warning method and system based on controllable cause

CN122017603ACN 122017603 ACN122017603 ACN 122017603ACN-122017603-A

Abstract

The invention discloses a battery fault early warning method and a system based on controllable causes, wherein the battery fault early warning method acquires multi-element time sequence data of batteries in different time periods by acquiring physical parameters of the batteries in the running process; after preprocessing the multi-element time series data of the battery, constructing a controllable factor vector according to the preprocessed data. The battery fault early warning model comprising the generator and the discriminator is constructed, the generator and the discriminator take the long-term memory network as a core, and the dependency relationship of the battery data on a long-time span can be effectively captured by introducing the long-term memory network into the generator, so that the accuracy of simulating the battery data is improved. And inputting the simulated data sequence by using the causal vector and the random noise into a generator, inputting the simulated data sequence and the real data sequence into a discriminator for countermeasure training, and obtaining a final discriminator model. And predicting the fault occurrence probability of the tested battery by using the final discriminant model.

Inventors

HE YIDONG
FANG YINFENG

Assignees

杭州芯锐特科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260407

Claims (10)

1. A battery fault early warning method based on controllable causes is characterized by comprising the following steps: Acquiring a data set containing different battery physical parameters, preprocessing the data in the data set, and acquiring a battery data sequence; respectively acquiring physical cause and state attribute of the battery according to the battery data sequence, and constructing a controllable cause vector based on the physical cause and the state attribute; The method comprises the steps of constructing a battery fault early warning model, wherein the battery fault early warning model comprises a generator G and a discriminator D, the generator G is used for mapping a random noise vector and a controllable factor vector into a simulated battery data sequence together, and the discriminator D is used for discriminating whether the battery data sequence is matched with the controllable factor vector corresponding to the battery data sequence; And training a battery fault early warning model by using different battery data sequences and corresponding controllable cause vectors, and carrying out fault early warning on the tested battery by using a trained discriminator.
2. The battery fault early warning method based on the controllable cause according to claim 1, wherein the physical cause comprises an operation state component, a health state component, an operation rate component and an ambient temperature component, and the state attribute adopts a fault type component.
3. The battery fault early warning method based on the controllable cause is characterized in that the method for acquiring the operating state component is characterized in that the operating state of a battery is judged based on the current, independent heat encoding is carried out on the operating state to acquire the operating state component, the health state component is the ratio of the current maximum available capacity to the rated capacity of the battery, the operating rate component is the ratio of average current to the rated capacity, and the environment temperature component is acquired through normalization processing of the operating temperature of the battery.
4. The battery fault early warning method based on the controllable cause according to claim 1, wherein the training process of the battery fault early warning model is as follows: The method comprises the steps of selecting a part of battery data sequence and corresponding controllable factor vectors to form a real data batch, randomly sampling noise vectors, sampling controllable factor vectors corresponding to the noise vectors one by one from the distribution of the controllable factor vectors, inputting different noise vectors and corresponding controllable factor vectors into a generator to obtain a simulated battery data sequence, constructing a simulated data batch based on the simulated battery data sequence and the corresponding controllable factor vectors, inputting the real data batch and the simulated data batch into a discriminator, acquiring a loss function L D of the discriminator according to a discrimination result of the discriminator, and updating network weight of the discriminator based on the loss function L D ; The method comprises the steps of obtaining random noise vectors through sampling, constructing corresponding controllable factor vectors, inputting different noise vectors and corresponding simulated controllable factor vectors into a generator G to obtain a simulated battery data sequence, inputting the simulated battery data sequence and the corresponding controllable factor vectors into a discriminator, and obtaining a loss function L G of the generator according to a discrimination result of the discriminator; And repeating the process until reaching the training termination condition, and finishing the training of the generator and the discriminator in the battery fault early warning model.
5. The battery fault early warning method based on the controllable cause according to claim 1, wherein the state attributes of the controllable cause vectors are all normal states.
6. The battery fault early warning method based on the controllable cause is characterized in that the generator comprises an input layer, a middle layer and an output layer which are sequentially connected, the input layer is used for splicing random noise vectors and the controllable cause vectors into fusion vectors and inputting the fusion vectors into the middle layer, the fusion vectors are processed in the middle layer sequentially through one or more full-connection layers and a remolding layer, the processing results are subjected to nonlinear transformation of time sequence characteristics through a long-period memory network to obtain an output sequence of the middle layer, and the output layer is used for processing the output sequence of the middle layer by adopting an activation function to generate a simulated battery data sequence.
7. The battery fault early warning method based on the controllable cause is characterized in that the discriminator comprises an input layer, an intermediate layer and an output layer which are sequentially connected, the input layer is used for splicing a controllable cause vector in a characteristic dimension with a battery data sequence after broadcasting in a time dimension to obtain a fusion sequence, the intermediate layer is used for processing the fusion sequence through a long-short-term memory network to obtain a concentrated representation of the whole sequence, and the output layer is used for processing the concentrated representation sequentially through one or more full-connection layers and an activation function to obtain a discrimination result of the discriminator.
8. The method for battery fault pre-warning based on controllable causes of claim 1, wherein the battery physical parameters include total voltage, total current, maximum temperature, minimum temperature, state of charge and state of health.
9. The battery fault early warning method based on the controllable cause according to claim 1, wherein the preprocessing method is characterized in that an abnormal value is identified and removed by adopting a quartile range method, and then data filling is carried out on the missing value by adopting one or more of a linear interpolation method, a forward filling method and a backward filling method.
10. The battery fault early warning system based on the controllable cause is characterized by being used for executing the battery fault early warning method based on the controllable cause, and comprises a battery data acquisition module, a data preprocessing module, a data enhancement module and a battery fault early warning module, wherein the battery data acquisition module is used for acquiring battery operation physical parameters to acquire battery data sequences of different time periods, the data preprocessing module is used for preprocessing the battery data sequences, the data enhancement module is used for constructing controllable cause vectors according to the battery data sequences, and the battery fault early warning module is used for predicting the occurrence probability of battery faults according to the battery data sequences and the controllable cause vectors.

Description

Battery fault early warning method and system based on controllable cause Technical Field The invention belongs to the technical field of battery fault early warning, and particularly relates to a battery fault early warning method and system based on controllable causes. Background In recent years, with the rapid development of new energy technology and computer science, battery safety early warning technology is receiving more and more attention. Lithium ion batteries are a common energy storage device characterized by high energy density but also associated with thermal runaway and other safety risks. Although the failure mechanism of the battery is complex and various, the safety of the system can be obviously improved through effective battery failure prediction, and the damage caused by accidents is reduced. Currently, research into battery fault pre-warning is mainly focused on analysis of voltage, current and temperature (BMS) signals, because these signals can directly reflect the electrochemical state of a battery and often show abnormal changes before a battery fault occurs. Traditional battery early warning methods are mostly based on fixed safety threshold values or simple machine learning algorithms, and the methods are often poor in effect when processing complex battery working conditions and early weak fault characteristics, so that the requirements of practical application are difficult to meet. Therefore, how to extract effective fault precursor features from complex BMS time sequence signals, solve the problem of scarcity of severe fault samples, improve the accuracy of battery fault early warning, and utilize advanced deep learning algorithm to perform intelligent battery early warning becomes the hot spot and difficulty of current research. Disclosure of Invention In order to solve the technical problems, the invention aims to provide a battery fault early warning method and system based on controllable factors, so as to solve the technical problems of detecting and identifying battery faults based on battery time sequence signals. In a first aspect, the present invention provides a battery fault early warning method based on controllable causes, the method comprising: Acquiring a data set containing different battery physical parameters, preprocessing the data in the data set, and acquiring a battery data sequence; respectively acquiring physical cause and state attribute of the battery according to the battery data sequence, and constructing a controllable cause vector based on the physical cause and the state attribute; The method comprises the steps of constructing a battery fault early warning model, wherein the battery fault early warning model comprises a generator G and a discriminator D, the generator G is used for mapping a random noise vector and a controllable factor vector into a simulated battery data sequence together, and the discriminator D is used for discriminating whether the battery data sequence is matched with the controllable factor vector corresponding to the battery data sequence; And training a battery fault early warning model by using different battery data sequences and corresponding controllable cause vectors, and carrying out fault early warning on the tested battery by using a trained discriminator. Preferably, the physical causes comprise an operation state component, a health state component, an operation rate component and an environment temperature component, and the state attribute adopts a fault type component. Preferably, the method for acquiring the running state component comprises the steps of judging the running state of the battery based on the current, performing independent heat coding on the running state to acquire the running state component, wherein the health state component is the ratio of the current maximum available capacity to the rated capacity of the battery, the running multiplying power component is the ratio of the average current to the rated capacity, and the environment temperature component is acquired through normalization processing on the working temperature of the battery. Preferably, the training process of the battery fault early warning model is as follows: The method comprises the steps of selecting a part of battery data sequence and corresponding controllable factor vectors to form a real data batch, randomly sampling noise vectors, sampling controllable factor vectors corresponding to the noise vectors one by one from the distribution of the controllable factor vectors, inputting different noise vectors and corresponding controllable factor vectors into a generator to obtain a simulated battery data sequence, constructing a simulated data batch based on the simulated battery data sequence and the corresponding controllable factor vectors, inputting the real data batch and the simulated data batch into a discriminator, acquiring a loss function L D of the discriminator according to a discrimination result of th