Remaining useful life prediction for mechanical equipment based on Temporal convolutional network

Abstract

For remaining useful life (RUL) prediction plays a very important part in prognostic and health management (PHM), How to improve the accuracy of remaining useful life prediction has been paid more and more attention by researchers. In recent years, the deep learning methods, especially the long-short term memory networks (LSTM), has proven to be excellent in fully excavating the time-dependent features of time-series data. However, recent studies have pointed out that the convolutional neural network should replace recurrent neural network as the first choice for processing sequence tasks and proposed a temporal convolutional network (TCN). In this paper, we proposed a remaining useful life prediction method based on temporal convolutional network (TCN). Firstly, the K-means clustering algorithm is used to identify the operating conditions of the system, the data is preprocessed under the same conditions. Then use sliding time window constructing the subsequence as the input of model. Finally, the prediction results of the proposed method and other advanced deep learning methods are compared on the public dataset C-MAPSS. Compared with other remaining useful life prediction methods, the method we proposed has higher prediction accuracy.