The Construction of Health Indicators and the Remaining Useful Life Prediction Methods for Bearings Through CNN-LSTM Network

This paper proposes a rolling bearing remaining useful life (RUL) prediction method based on multi-feature factor analysis and dynamic window correction. Effective health indicators are extracted through factor analysis, the elbow point detection method is used to determine the first prediction time, and dynamic window correction is applied to reduce the impact of random fluctuations. By combining the exponential degradation model with the CNN-LSTM network, accurate RUL predictions are achieved. Compared with other methods, this approach performs excellently on open-source datasets and effectively predicts the health state and degradation trend of rolling bearings.