Fault Diagnosis of TE Process Using LSTM-RNN Neural Network and BP Model

the classification accuracy of Tennessee Eastman (TE) chemical process fault diagnosis is low. In this study, a Long short recurrent neural network (lstm-rnn) model is proposed, which can effectively improve the defects of RNN recurrent neural network that gradients disappear and explode easily with time. Finally, the results of lstm-rnn model, BP model and RNN model are compared to verify the advantages of this method. It is found that lstm-rnn model has stable classification error and high classification accuracy, which effectively improves the fault diagnosis ability of te chemical process.