Interpretable Model for Remaining Useful Life Prediction of Batteries

With the rapid advancement of renewable energy and the widespread adoption of microgrids, there is a growing need to predict the remaining lifespan of microgrid batteries. However, conventional prediction methods based on traditional artificial intelligence models suffer from a lack of interpretability, which hampers their reliability and credibility in practical applications. Addressing this concern, this research paper proposes a method based on interpretable artificial intelligence models for accurately predicting the remaining lifespan of microgrid batteries. To begin with, we collected charging and discharging datasets consisting of lithium batteries and performed preprocessing and feature extraction. In addition to the classic linear prediction model, we incorporated an interpretable module. This module enables the automatic learning of complex battery data features, generating interpretable outputs while maintaining highprecision predictive performance. To validate the effectiveness of our proposed method, multiple sets of simulation experiments were conducted. The results demonstrate that our model can accurately and interpretably predict outcomes, providing valuable insights for the maintenance and management of microgrid batteries.