(583b) Fault Detection and Diagnosis of Chemical Processes by An Immune-System Inspired Approach

Quick and correct detection and identification of process faults are extremely important when efficient, economic and safe operation of chemical processes is concerned. Undetected process fault may lead to poor quality off-spec products, resulting in poor plant economy and sometimes even catastrophic consequences like accidents, injury to plant personnel. Successful detection and identification of process faults at an early stage can increase the success rate of fault recovery during operations and prevent costly accidents, unnecessary shutdowns. Detection and diagnosis of process faults in chemical processes has been an active area of research. In the literature, several methodologies have been proposed for fault detection and identification (FDI) in chemical processes (Venkatasubramanian et al. 2003 a,b,c; Dash et al. 2000, Chiang et al. 2001) including principal components analysis (PCA), artificial neural-networks (ANN), self-organizing maps (SOM), qualitative trend analysis (QTA), signal processing methods or first principles models. Each of these methods has its advantage and weakness in practical application. To overcome the limitations of an individual method, one needs to develop a system in which multiple FDI methods are judiciously combined (fused) or new fault detection and identification (FDI) systems.

Artificial immune system is a new artificial intelligence methodology that is increasingly attracting much attention for monitoring engineered systems. In an artificial immune system (AIS), principles and processes of the natural immune system are abstracted and applied in pattern recognition, anomaly/ novelty detection, computer security, machine learning and a variety of other applications in the field of science and engineering. One popular immune-inspired approach is the Negative Selection Algorithm (NSA). The negative selection is a process in natural immune system in which self-tolerant T-cell are generated, thus allowing the immune system to discriminate self proteins from foreign (non-self) ones. This principle inspired Forrest et al. (1994) to propose a negative selection algorithm to detect data manipulation caused by computer viruses. The basic idea was to generate a collection of detectors, usually called detector set in the complementary (non-self) space and then to apply these detectors to classify new (unseen) data as self or non-self. It is typically regarded as an anomaly detection or a one-class classification method because the training data are from normal cases (self samples) only. It is now used extensively in anomaly detection, intrusion detection, novelty detection, fault detection particularly in the situations where only large amount of self (normal) samples are available but abnormal samples are either unavailable or very rare.

In this work, we propose a real-valued NSA based framework for Fault detection and identification (FDI) of chemical process in real time. The effectiveness of the proposed real-valued NSA based FDI framework is demonstrated through the online fault detection and diagnosis in two case studies ? a continuous lab-scale distillation column and a CSTR. The results show that the proposed NSA based framework provides excellent monitoring and diagnosis performances for both these cases with (i) complete fault coverage ? all the faults studied can be readily detected and identified, (ii) high overall recognition rate (~90%), (ii) low false positive rate (~<3%), (iii) high true positive rate (~>95%), (iv) early fault detection (~10-15 samples delay) and diagnosis (~60-80 samples delay). A comparative study with other data-driven approaches will also be reported.

References:

Venkatasubramanian, V., Rengaswamy, R., Yin, K., Kavuri, S.N., (2003a). A review of process fault detection and diagnosis Part I: Quantitative model-based methods, Computers and Chemical Engineering 27, pp. 293 ? 311.

Venkatasubramanian, V., Rengaswamy, R., Kavuri, S.N., (2003b) A review of process fault detection and diagnosis Part II: Qualitative models and search strategies, Computers and Chemical Engineering 27, pp. 313 ? 326.

Venkatasubramanian, V., Rengaswamy, R., Kavuri, S.N., Yin, K., (2003c). A review of process fault detection and diagnosis Part III: Process history based methods, Computers and Chemical Engineering 27, pp. 327 ? 346.

Dash S. and Venkatasubramanian, V., (2000). Challenges in the industrial applications of fault diagnostic systems, Computers and Chemical Engineering 24, pp. 785 ? 791.

Chiang, L.H., Russell, E.L., Braatz, R.D., (2001). Fault detection and diagnosis in industrial systems, Springer-Verlag London Limited, Briton.

Forrest, S., Perelson, A., Allen, L., R., and Cherukuri (1994). Self-nonself discrimination in a computer. In Proceedings of the 1994 IEEE Symposium on Research in Security and Privacy, pages 202?212, Los Alamitos, CA. IEEE Computer Society Press.