ABSTRACT
Nowadays, global industry is witnessing the explosion of exible manufacturing systems considered as key role for the fourth industrial revolution. Almost hardware and software producers tried their best to assure the best performance for those system. However, some chemical processes in particular requires more strict conditions beside high-accuracy operation of actuators. Chemical reactions may lead to a poor quality for output product even a small change. Therefore, beside the necessary of high-performance integrated control system, monitoring operation should be fast and correct enough to detect and isolate faults when any problems happened in system. In this paper, a method for process fault detection and diagnostic based on data-driven estimation is investigated. In this method, the process fault is detected based on the error between process model response and process response. The Kernel Principal Component Analysis (KPCA) is utilized to classify errors for fault diagnostic. In this research, the method is veried on Stirred-Tank Heating Process. The simulation results demonstrate the effectiveness of proposed method