431329 Improvement of Artificial Immune System in Fault Detection and Diagnosis of Chemical Process

Monday, November 9, 2015
Exhibit Hall 1 (Salt Palace Convention Center)
Liang Ming, department of chemical engineering, tsinghua university, Beijing, China and Jinsong Zhao, Responsible Production and APELL Center (UNEP), Department of Chemical Engineering, Tsinghua University, Beijing, 100084, China, Beijing, China

Because of the self-adaptation and self-learning ability of artificial immune system(AIS), it is introduced to fault detection and diagnosis of chemical process. In this paper, antibodies and antigens are composed of matrices of time-sameple data of process variables, and normal antibody libraries and fault antibody libraries are later constructed based on clonal selection algorithms. Since chemical process variables are always of a great number and change frequently and irregularly, matrix of each antibody is only composed of variables of the root cause of the corresponding fault, which helps to avoid the interference of changes of other process variables, improve the prediction and accuracy of AIS and save computing time and storage. To meet the requirement of adequate historical data of different kinds of faults, antibody libraries are transplanted among chemical processes of the same kind.

Extended Abstract: File Not Uploaded