期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2021
卷号：48
期号：4
语种：English
出版社：IAENG - International Association of Engineers
摘要：In the studies of machinery remaining useful life (RUL) prediction, the construction of health indicator (HI) can appropriately describe operating condition of machinery, which is always the bridge between raw data and RUL. In this paper, an improved HI construction method is utilized to solve glitches introducing when outlier region correction, which are adaptive generalized framework of HI construction and can apply to vibration-signal-related scenarios. On the basis of it, gated recurrent units (GRUs) are proposed to investigate the degradation information over time aim to output a estimated RUL as close to actual RUL as possible. The proposed method in this paper is compared against the state-of-the-art. For further verification, we have choose two dataset come from varying platform when experimental verification. At the same time, long short-term memory (LSTM) also is the one of baseline, which makes recurrent-neural-networks-based (RNN-based) architectures identified as an effective method for RUL prediction. Experimental verification has been carried out on FEMTO and XJTUSY dataset, experimental results testify proposed method outshines other baselines.
关键词：PHM;health indicator;remaining useful life prediction;gated recurrent units
