Detection and Isolation of Abnormal Event in Nonlinear Industrial Processes By a Novel Data-Based Method

Detection and Isolation of Abnormal Event in Nonlinear Industrial Processes by a Novel Data-Based Method

Chiranjivi Botre^a, Hazem N. Nounou^b, Mohamed N. Nounou^c, and M. Nazmul Karim^a

^a Artie McFerrin Dept. of Chemical Engineering, Texas A&M University, College Station, Texas 77843, USA

^b Electrical and Computer Engineering Program, Texas A&M University at Qatar, Doha, QATAR,

^c Chemical Engineering Department, Texas A&M University at Qatar, Doha, QATAR

Abstract

Detection and isolation of abnormal events are important aspects in the industry to ensure safe and proper operation of the plant. It is important to accurately, reliably, and quickly detect changes or faults within the process, determine the cause of these changes, and apply corrective actions to mitigate risk. Multivariate statistical techniques are powerful tools that utilize data based model reduction methods capable of efficiently handling process noise and correlated data sets. In literature, multivariate statistical methods are widely discussed for applications in process monitoring and fault detection. Moreover, data based process monitoring techniques have been successfully applied to applications where the accurate process model are not available. Fault detection can be carried out in two phases; data based model reduction and statistical fault detection.

The partial least square (PLS) method is an input-output model that has been effectively applied to linear processes and is most commonly used for fault detection applications. A kernel extension of PLS has been proposed to provide an effective technique for modeling nonlinear industrial processes. In order to enhance the fault detection performance of KPLS model we have developed a new method that optimizes KPLS model through the use of a multi-objective genetic optimization approach. Our multi-scale kernel partial least square (MSKPLS) - based generalized likelihood ratio test (GLRT) method, has the ability to handle process noise, non-normal data distribution, and auto-correlated data sets to provide an effective fault detection technique that can be applied to nonlinear industrial process data.

Once process faults have been detected, accurate isolation of the specific fault within the plant must be achieved to aid swift corrective action. Our approach to fault detection using the advanced statistical GLRT method also allows for multiple faults in the process to be isolated. The isolation method is called a contribution plot and uses the calculated statistical values to determine specific process variables responsible for the process fault, narrowing the possible fault location from plant-wide consideration to the contributions of a key sub-block consisting of a few unit operations. We have developed a novel fault isolation and identification hybrid approach by using a targeted approach where the generalized likelihood ratio (GLRT) based contribution plot is used to isolate the faulty sub-block of a process, and then hybrid model-based observer with a neural network is used to isolate and identify the faulty equipment. We will demonstrate the improved fault detection and isolation algorithm using the nonlinear simulated continuously stirred tank reactor (CSTR) data and Tennessee Eastman Process problem as a case study.

Keywords: KPLS, Wavelet function, Hybrid observers, Fault isolation, Neural network, Continuously stirred tank reactor (CSTR).

References:

1.       Bakshi, B. R. Multiscale PCA with application to multivariate statistical process monitoring. AIChE J. 44, 1596–1610 (1998).

2.       Botre, Chiranjivi, Majdi Mansouri, Mohamed Nounou, Hazem Nounou, and M. Nazmul Karim. 2016. “Kernel PLS-Based GLRT Method for Fault Detection of Chemical Processes.” Journal of Loss Prevention in the Process Industries 43 (September): 212–24.

3.       Botre, C., Mansouri, M., Karim, M. N., Nounou, H. & Nounou, M. Multiscale PLS-based GLRT for fault detection of chemical processes. J. Loss Prev. Process Ind. 46, 143–153 (2017).

4.       MacGregor, J.F., and T. Kourti. “Statistical Process Control of Multivariate Processes.” Control Engineering Practice 3, no. 3 (March 1995): 403–14.

5.       Teppola, Pekka, and Pentti Minkkinen. 2000. “Wavelet–PLS Regression Models for Both Exploratory Data Analysis and Process Monitoring.” Journal of Chemometrics 14 (5-6): 383–99.