Agent Based Fault Detection System for Chemical Processes using Negative Selection Algorithm

Abstract

Recently, the number of industrial accidents of chemical plants has been increasing in Japan. The fault detection system is required to keep chemical plant safely. In this study, a fault detection system for a chemical plants using agent framework and negative selection algorithm was proposed. The negative selection algorithm is one of artificial immune systems. The artificial immune system is an imitative mechanism of vital actions to discriminate self/nonself to protect itself. The method was implemented and applied to a complicated chemical plant-which is a boiler plant virtually operated using a dynamic plant simulator. The simulations of fault detection were carried out. And also the results of simulations are presented in this paper.