Battery anomaly detection model training and detection method and device
An anomaly detection and model training technology, applied in the direction of measurement devices, neural learning methods, biological neural network models, etc., can solve problems such as inconvenient positioning, poor interpretability of machine learning models, and inability to explain the reasons for battery abnormalities well , to achieve the effect of improving interpretability
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
no. 1 example
[0030] The first embodiment of the present application provides a battery abnormality detection model training method, figure 1 The main flow of the battery abnormality detection model training method of this embodiment is shown, as shown in the figure, it includes the following steps:
[0031] Step S11, building a battery abnormality detection model.
[0032] In this embodiment, the constructed battery abnormality detection model includes an encoder, a decoder and a classifier.
[0033] Optionally, the battery abnormality detection model constructed in this application is a deep generative model trained based on a meta-learning strategy, for example, a variational autoencoder model (hereinafter referred to as a VAE model).
[0034] Such as Figure 2A As shown, in this embodiment, the encoder in the constructed battery abnormality detection model includes an input layer, at least one CNN layer, a Flatten layer, an LSTM layer and a fully connected layer arranged in sequence. ...
no. 2 example
[0062] The second embodiment of the present application provides a battery abnormality detection model training method, image 3 The main flow of the battery abnormality detection model training method of the embodiment of the present application is shown, as shown in the figure, it includes the following steps:
[0063] Step S31, for each of the plurality of cells, obtain multiple behavior data corresponding to each cell according to the preset time step;
[0064] In this embodiment, the preset time step can be between 1 minute and 10 minutes, but it is not limited thereto, and can be adjusted according to actual needs.
[0065] Step S32, performing first pre-processing on the behavior data acquired by the cells based on the first preset data processing rule, so that the dimensions of the behavior data corresponding to the different cells are the same.
[0066] Specifically, when the charging time of different battery packs is different, the amount of behavior data of each b...
no. 3 example
[0073] The third embodiment of the present application provides a battery abnormality detection model training method, which mainly shows the specific implementation steps of step S13, as shown in Figure 4 As shown, the battery abnormality detection model training method of this embodiment mainly includes the following steps:
[0074] Step S41, using the encoder to generate mean vector parameters and variance vector parameters according to the first training parameters.
[0075] Specifically, the encoder can be used to perform encoding processing according to the input first training parameters to obtain the hidden layer features of the VAE model, that is, the mean vector parameter and the variance vector parameter
[0076] In step S42, a decoder training operation is performed, and the decoder is trained using the mean vector parameter and variance vector parameter generated by the encoder.
[0077] Specifically, the decoder is trained by using the generated mean vector par...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More