Static Friction Detection Based on Artificial Neural Networks Method

Abstract

Introduction: Poor product quality and high energy consumption ofmany control loops is due to the presence of static friction. This phenomenon is monitored by human in many industrials. The decision ismade based on human’s brain which is not effective and reliable. Methods: A model-based method of stiction detection based on an artificialneural network (ANN) is proposed. The ANN which is run in parallel tothe process predicts a dynamic model of the process using data obtainedfrom control signal and process output. Results: It can be seen that theproposed method based on ANN can be replaced with human monitoring method. Conclusions: Capability of the proposed method of staticfriction detection for the process with the sticky valve is confirmed bydata obtained from the simulation in a control loop with sticky valve.