A Segmented Method for Predicting the Remaining Life of Bearings Using LSTM
作者
X. H. Li,Meidi Sun,Tangyan Chen,Qing He
标识
DOI:10.1109/rcae66389.2025.11355199
摘要
Rolling bearings are key components in mechanical systems. Failure of bearings lead to equipment downtime, safety accidents, and significant economic losses. Predicting the remaining life of a bearing can improve equipment reliability, reduce maintenance costs, and increase productivity. A segmented method for predicting the remaining life of bearings using Long Short-Term memory network (LSTM) is proposed in this paper to improve the prediction accuracy. First, time domain, frequency domain and time-frequency features are extracted from the vibration signals, and effective features are screened by constructing comprehensive evaluation indexes. Second, t-SNE algorithm and DBSCAN clustering are used to adaptively classify bearing life into healthy and degraded stages. Finally, a segmented LSTM model is established for predicting RUL at different stages. The experiment is based on the bearing dataset from Xi'an Jiaotong University. The results of the comparative experiments show that the average absolute error of the prediction after stages is reduced by 5.12% compared with the traditional method, which verifies the effectiveness of the proposed method.