Remaining Useful Life Prediction Method Based on the Improved Holt Double Exponential Model

Abstract

For the problem of low prediction accuracy due to the difficulty in determining the parameters of Holt double exponential model, a remaining useful life (RUL) prediction method based on deep residual shrinkage network (DRSN) and Holt double exponential model is proposed. Firstly, without any human experience interference, the DRSN model is used to self-extract features from the original signal of equipment condition monitoring, mine degradation features, and construct health indicators; then, the Holt double exponential model is used to construct the remaining life prediction model, and the sparrow search algorithm (SSA) is used to optimize the two parameters for the problem that the Holt double exponential prediction model hyperparameters , are difficult to determine. After smoothing the health indicators extracted by DRSN, the optimized Holt double exponential prediction model is used for remaining useful life prediction. The proposed method was verified by comparative experiments using the experimental data of the whole life of the bearings. The experimental results show that the proposed method in this paper effectively improves the remaining useful life prediction accuracy and provides a new way of thinking for predictive maintenance of equipment.