A Fault Detection Method based on Convolutional Gated Recurrent Unit Auto-encoder for Tennessee Eastman Process

Due to the powerful ability of feature learning, deep learning has been widely used in the field of manufacturing process monitoring. This paper proposes a new fault detection method, convolutional gated recurrent unit auto-encoder (CGRU-AE) for feature learning from process signals. CGRU-AE is used to learn the effective features of time series data for fault detection. Two statistics, the T² and the squared prediction error (SPE), are generated in the feature space and the residual space of CGRU-AE, respectively. Finally, the feasibility and advantages of CGRU-AE are shown on the Tennessee-Eastman process. The results show that CGRU-AE is capable of learning effective features from complex process variables and outperforms other typical methods.