Remaining Useful Life Prediction for Pneumatic Control Valve System Based on Hybrid CNN-LSTM Model

Abstract

Pneumatic control valve is an important terminal control element in the industrial control system. Accurate prediction of the residual life of pneumatic control valve is of great significance to industrial production. This paper introduces an innovative fault diagnosis and remaining useful life (RUL) prediction method for a pneumatic control valves system. Firstly, the parameters with relatively large correlation coefficients are selected as inputs using Perason’s law. Next, support vector machine (SVM) is used to classify the fault types. Then, convolutional neural network (CNN) is employed to capture local features from the input data, and long short term memory (LSTM) is used to extract long-term dependencies from the features. Finally, fully connected networks are used to predict the RUL. Through experimental verification, the prediction model proposed in this paper has better prediction performance than other proposed models.