Skip to Main Content
A new ensemble algorithm for support vector machine (SVM) based on attributes reduction and parameters disturbance is proposed, which is applied to analog circuit fault diagnosis. Firstly, the feature space is divided in several subspaces by attributes reduction algorithm with assurance of high classification capability. And then, for each subspace, the model parameters are disturbed in “low-bias region”. The final result is obtained by using the majority voting procedure twice. Take Sallen-key band-pass filter as simulation instance, and the fault diagnosis result indicates that the algorithm presented in the paper has better performance then single SVM, Adaboost algorithm, “Attribute Bagging” algorithm and so on.