Fault diagnosis of batch process based on improved time convolution network and efficient channel attention

Abstract

Aiming at the nonlinear and non-gaussian characteristics of batch processes, a fault diagnosis model for batch processes according to the improved time convolution network(TCN) and efficient channel attention(ECA) is proposed. Standardize the 3D data, and then input the standardized data into the model combined with the time convolution network of hybrid dilated convolution and efficient channel attention to extract features. Finally, use the softmax function to output the fault diagnosis tag. The excellence of the proposed model is verified by the simulation of penicillin experimental data and the comparison with the classical depth learning method.