Test entrance´s selection for fault diagnosis of a self-propelled fire control system based on TC-SVM

Due to the complexity of the self-propelled fire control system and the difficulty of test entrance´s selection, the support vector machine combined with text categorization method (TC-SVM) is applied as the first step of fault diagnosis. The text pretreatment of text representation and feature extraction are introduced first. Then the error-correcting code is used to reduce the multiclass classifier to several binary classifiers which are performed by the support vector machine. Finally, the methods of how to combine the output of binary classifiers as the final diagnosis result is discussed, two improved algorithms are put forward based on the basic decoding method.