A Remaining Useful Life Prediction Method for Rolling Bearings Based on Broad Learning System - Multi-Scale Temporal Convolutional Network

Rolling bearings play a critical role in various industrial applications.However, the complexity and diversity of data, along with the challenge of selecting the most representative features from a large set and reducing dimensionality to lower computational costs, pose significant challenges for accurately predicting the remaining useful life (RUL) of rolling bearings.To address this, a hybrid model combining the broad learning system (BLS) and multi-scale temporal convolutional network (MsTCN) is proposed for RUL prediction of rolling bearings.The BLS is employed to capture a broad range of features from the full-life signals of rolling bearings, while the MsTCN adaptively extracts multi-scale temporal features, effectively capturing both short-term and long-term dependencies in the bearing's operational process.Additionally, the fusion and optimization of features extracted by BLS and MsTCN enhance the representational power of the prediction model.Experiments conducted on the PHM2012 bearing dataset demonstrate that the proposed method significantly improves model performance and prediction accuracy.