Abstract:A novel semi-supervised Support vector machine fault detection model based on two stage learning is presented in this paper. First of all, propagation algorithm is applied to provide pseudo labels for the unlabeled samples. Secondly k-near-neighbor graph is utilized to distinguish and delete the noisy samples. Then the denoised samples are input into the support vector machine(SVM), so that the global information of the whole samples can be considered by SVM to enhance the fault detection accuracy. The comparisons with Support vector machine, Fuzzy Support vector machine, Transductive support vector machine and Laplacian support vector machine fault detetion algorithm are performed.The experiments show that the proposed approach has a more substantial detection accury than other algorithms under different labeled sample sets. The proposed fault detection methods with test samples and without test samples are compared.The results illustrate the investigated techniques with test samples as unlabeled samples can outperform the one without test samples as unlabeled samples.