Optimum selection of haar type orthogonal matrix in the fault diagnosis

Haar type orthogonal matrices (HTOMs) include Haar, Walsh and other orthogonal matrices intervenient former two. These matrices not only have fast algorithm but also can be obtained conveniently by varying any one of two parameters in the same program. However, they are rarely utilized in practice. For this reason, HTOMs are applied in the fault diagnosis in this paper. Namely, different mechanic state signals are transformed by various HTOMs, respectively. Meanwhile, Fisher classification rule is viewed as the evaluation criterion, and the transform results are compared. As a result, the most optimal HTOM in the fault diagnosis is obtained.