Performance Evaluation Of Statistical Method For Incipient Fault Detection Under Noisy Environment

Dealing with incipient fault detection and diagnosis particularly in noisy environment, practical and reliable solutions are the major challenge in the industrial process. In this paper, analytical studies have been performed for the fault detection and performance characterization using false alarm and missed detection probabilities considering the noisy environment. Proposed model-based method combines the optimal filter with Generalized Likelihood Ratio (GLR) test to cancel out fault dynamics and has proven to be particularly efficient for incipient fault detection and identification. However, performance of the technique is highly dependent on the chosen threshold, Signal to Noise Ratio (SNR) and Fault to Noise Ratio (FNR). To illustrate the effectiveness of proposed method, an incipient fault in a temperature sensor of benchmark CSTR process is considered. (C) 2017, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.