Fault classifier of rotating machinery based on weighted support vector data description

This paper presents a novel fuzzy classifier for fault diagnosis of rolling machinery based on support vector data description (SVDD) and kernel possibilistic c-means clustering. The proposed method considers the effect of negative samples, which should be rejected by positive class, to the SVDD classifier. Firstly, we compute weights of training samples to the given positive class using the kernel PCM algorithm. Then according to weights, we select some meaning samples to construct a new training set, and train these samples with the proposed weighted SVDD algorithm. The proposed method is applied to the fault diagnosis of rolling element bearings, and experimental results show that the proposed method can reliably separate different fault conditions, and reduce the effect of outliers to classification results.