Research on Remaining Lifetime Prediction of IGBT Using LSTM Neural Network Optimized by an Improved Snake Optimization Algorithm

This paper proposes a novel method for predicting the remaining lifetime of IGBT based on the publicly available NASA dataset. This approach utilizes the collector-emitter blocking voltage as a failure parameter and constructs a prediction model utilizing a Long Short-Term Memory (LSTM) neural network. To enhance the prediction performance, the snake optimization algorithm is introduced, and an improved method is proposed, which incorporates a K-means++ initialization strategy to initialize the population and introduces a non-uniform mutation strategy to address its limitations. Experimental results demonstrate that the improved model achieves high accuracy in predicting the remaining lifetime of IGBT, outperforming the other three comparison models. This study provides a new perspective for the field of IGBT remaining lifetime prediction, offers valuable insights for the stable operation of power electronic systems, and looks forward to further exploring and optimizing this model in future work.