Abstract: Support vector data description (SVDD) is a kind of one-class classification method. It can be used to solve the problem of insufficient fault samples in fault diagnosis. Vector-bispectrum is the bispectrum analysis method based on the full vector spectrum information fusion. It can be used to fuse the double-channel information of the rotary machines effectively and reflect the nonlinear properties in the signals more completely and accurately. In order to realize the aim that the faults of the machines can be diagnosed effectually and intelligently under the situation of lack of fault samples, an intelligent fault diagnosis method by combining vector-bispectrum with SVDD is put forward. By using the vector-bispectrum to process the signals and extract the characteristic vectors to be used as the input parameters of SVDD. The classification model is set up and therefore the running states of the machines can also be classified. The method is applied to the gearbox fault diagnosis. The results indicate that the method can be effectively used to extract the characteristic information of the gearbox signals and increase the accuracy of SVDD in the fault diagnosis.

Key words: Full vector spectrum, Information fusion, Intelligent diagnosis, Support vector data description, Vector-bispectrum