A Remaining Useful Life Prediction Method for Insulated-Gate Bipolar Transistor Based on Deep Fusion of Nonlinear Features From Multisource Data

Abstract

An insulated-gate bipolar transistor (IGBT) has multiple degradation mechanisms; it is a challenge to accurately integrating multiple signals to capture the device’s degradation patterns and health state. Therefore, comprehensively characterizing the health state of IGBT and predicting its remaining useful life (RUL) using multiple signals poses a significant challenge. To address this challenge, a RUL prediction method for IGBT based on the deep fusion of nonlinear features from multisource data is proposed. First, the time-domain multifeatures of IGBT degradation data are constructed, and key features are selectively selected; then, dimensionality reduction is performed and these features are fused into health indicators (HIs) to characterize the health level. Second, the health of IGBT is effectively evaluated by unsupervised clustering without data labeling. Third, end-condition monitoring is refined to enable the identification of near-failure state. Finally, deep learning is utilized to provide the accurate and reliable prediction of the RUL of IGBT devices [ ${R}^{{2}}$ are all greater than 0.98, the mean absolute error (MAE) all less than 2.3, and the root mean square error (RMSE) all less than 5.5.]. The results demonstrate that the method effectively integrates multisource information, characterizes the health state of the device, and can more accurately and reliably predict the RUL of IGBT. The proposed method can enhance the scientific basis for the health management of new energy systems such as wind power and photovoltaic systems.