In order to solve the fault diagnosis of aircraft power starting system, the fault diagnosis method through constructing binary tree SVM (support vector machines) is researched in this paper. Consider that the components which have high fault rate have priority to be isolated, the method trains the classifier depend on MTBF from small to big and uses the basic structure of binary tree-SVM to generate the leaf nodes gradually, each leaf node represents a fault mode. The method is applied to the identify the common nine fault mode of aircraft power starting system.
展开▼
