FAULT DIAGNOSIS USING SYSTEM IDENTIFICATION FOR CHEMICAL PROCESS PLANT

Fault detection and diagnosis have gained an importance in the automation process industries over the past decade. This is due to several reasons; one of them being that sufficient amount of data is available from the process plants. The goal of this project is to develop such fault diagnosis systems, which use the input-output data of the realm process plant to detect, isolate, and reconstruct faults. The first part of this project focused on developing a different prediction models to the real system. Moreover, a linearized model using Taylor Series Expansion approach and ARX (Autoregressive with external input) model of the real system have been designed. In addition, the most accurate identification model which describes the dynamic behavior of the monitored system has been selected. Furthermore, a technique Statistical Process Control (SPC) used in fault diagnosis. This method depends on central limit theorem and used to detect faults by the analysis of the mismatch between the ARX model estimation and the process plant output. Finally the proposed methodology for fault diagnosis has been applied in numerical simulations to a non-isothermal CSTR (continuous stirred tank reactor) and the results and conclusion have been reported and showed excellent estimation of ARX model and good fault diagnosis performance of SPC.