Chemical Engineering Science, Vol.53, No.16, 2921-2931, 1998
Minimax methods for fault isolation in the directional residual approach
This work designs a robust fault isolation method applicable to the directional residual approach by using decision theory. The cost, the loss, and the risk of misclassification are analyzed. To achieve a balance between optimality and robustness, a minimax method is proposed which minimizes the maximum expected loss from misclassification. An isolation procedure is designed for the diagnostic residuals having normal distribution. A simulated distillation example is given to illustrate the implementation of this technique. The suggested approach is compared to the methods derived from the likelihood ratio tests and their relationship elucidated.
Keywords:GRAPHS PCEG MODEL;FAILURE-DETECTION;NEURAL NETWORKS;PROCESSPLANTS;DIAGNOSIS;SYSTEMS;REDUNDANCY;DESIGN