Fault diagnosis of plant systems using immune networks

Abstract

Recently, systems such as chemical and nuclear plant systems have been increasing in scale and complexity. In these systems, when a certain device (unit) in a plant system becomes faulty, its influence propagates through the whole system, and then causes a fatal situation. To construct the safety and reliability of plant systems, the necessity for an efficient fault diagnosis technique is increasingly demanded. On the other hand, biological systems such as human beings can be said to be the ultimate information processing system, and are expected to provide feasible ideas to engineering fields. Among the information processing systems in biological systems, immune systems work as online fault diagnosis systems by constructing large-scale networks, called immune networks (idiotypic networks). In this study, we try to apply these immune networks to a fault diagnosis of plant systems. And the feasibility of our proposed method is confirmed by simulations.<>