Research on Fault Classification Model of TE Chemical Process Based on LRCN

Due to the low efficiency and poor classification accuracy of existing fault classification methods in processing high-dimensional and large-scale chemical process data, a classification model combining sliding window and Long-Term Recurrent circular convolution (LRCN) networks is proposed in this paper. The simulation data set generated by the Tennessee Eastman (TE) simulation platform which based on the actual chemical reaction process is used as the research object to verify the model. The experimental results show that compared with the long short-term memory (LSTM) method, LRCN is more reliable, the model classification accuracy is higher, the training efficiency is faster, and it is more suitable for processing large data sets.