Chemical process fault diagnosis based on mixup-convolution neural network

Accidents in the chemical production process will have serious consequences, so the fault diagnosis system is extremely important. To increase the accuracy, it is a method to expand the size of the neural network, but the sensitivity of the test sample and the training sample is. In this paper, using the fault diagnosis method of mixup-convolution neural network (CNN) model can extract more abundant fault information from time-varying features. Mixup uses neighborhood data training model to have better generalization ability and can overcome large-scale network. The problem of remembering data. This experimental data uses TE process data for simulation experiments, and the final experimental results can verify the performance of the proposed method.