Rolling Process Failure Diagnosis Based on PCA-LSTM Neural Network
Aiming on the failure and the failure of cooling water valve,the complex data structure,influence and nonlinear char-acteristics.The extraction of the main component analysis(PCA)improves the iteration speed during the rolling process network training and reduces the input dimension and prediction difficulty of the LSTM network.Use the extracted feature data as the in-put to the LSTM network and the fault category as the output.Diagnosis of two types of faults using the PCA-LSTM troubleshoot-ing model.Comparing the diagnostic results of PCA-LSTM、LSRM、BP neural networks shows that the data training network iter-ation after using the PCA-LSTM diagnostic model is faster and the diagnostic accuracy exceeds 99％,with better diagnostic ef-fect over ordinary LSTM and BP neural networks.
万方数据
维普
