Abstract:

Aiming at the problem of multi-fault pattern recognition for uncertain data, a method based on support vector data description is proposed. The possibility measure is introduced into support vector data description (SVDD) as a constraint condition. First, the programming is transformed into its dual form. Then, a fast training method based on sequential minimal optimization (SMO) is developed to get the result. Finally, a multi-class SVDD model is established based on one-class SVDD. Numeric experiments show that the accuracy and speed of classification are improved, and this method is suitable for practical use.