Kernel least squares regression for fault isolation

In this article, a framework of discriminative kernel least squares regression (KLSR) is proposed for fault isolation. The core concept is to enlarge the distance between different classes of faults under the conceptual framework of KLSR. First, a technique called ε-dragging is introduced to force the regression targets of different classes of faults moving along opposite direction such that the distance between classes can be enlarged. Then ε-dragging is integrated into the KLSR model for fault isolation. The fault is finally isolated elegantly and efficiently. Experimental evaluation over a range of benchmark dataset indicates the validity of our method.