Remaining Useful Life Prognostics Based on Deep Combined Temporal Bidirectional Convolutional Network

The prognostics and health management (PHM) is the key technology to protect the security and ability of the machine, it has gotten remarkable theoretical achievement and have broaden application. The remaining useful life (RUL) is the important part of the PHM, which describe the time interval between the current time and fault time. Recent year, RUL based on deep learning has made rapidly development, such as deep convolution neural networks (DCNNs). This paper proposes a new data driven model for prediction using deep combined temporal convolution neural and bidirectional long short-term network. To make an evaluation about this new approach, the C-MAPSS dataset is used for testing the performance of the new model. The comparisons with other approaches also suggest the proposed new deep network offers a new and promising approach.