Research on performance degradation assessment method of train rolling bearings under incomplete data

This paper mainly discusses the performance degradation assessment of train rolling bearings under incomplete data, by using the support vector data description (SVDD) and dynamic particle swarm optimization (DPSO).The proposed method is based on the similarity weight for the assessment of the train rolling bearings under incomplete data. Firstly, to obtain effective features of bearing performance degradation from collected vibration data, the local mean decomposition (LMD) is employed to decompose the vibration data. Secondly, the high-dimensionality of features is reduced by the principal component analysis (PCA). And then, on the basis of choosing the kernel parameter and penalty weight, a degradation method based on SVDD is proposed. Finally, the experimental results verified that the proposed method has a better optimization performance than the traditional method and can assess the performance degradation of train rolling bearings under incomplete data.