Battery abnormality detection based on important sample selection for large-scale vehicle monitoring

With the continuous development of artificial intelligence technology, deep learning algorithms have been widely applied to abnormality detection of power battery in academia. However, in the face of large-scale real-world vehicle monitoring, there are realistic bottlenecks in deploying deep learning algorithms on massive training samples due to the limitations of factors such as time cost and computing resources. In order to solve the above problem, this paper proposes a battery abnormality detection method based on important sample selection. This approach conducts importance evaluation at the sample level, selecting important samples from the original training samples for model training, thus reducing training time while ensuring model performance. Data from twenty real-world vehicles collected over a two-year period was used for validation. The results indicate that the proposed method achieves comparable performance while reducing training time by up to 61 %, thereby providing basis for the deployment of deep learning-based battery abnormality detection methods in large-scale vehicle monitoring.