ROLLING BEARING FAULT DIAGNOSIS BASED ON UNIFORM DISTANCE LOCAL LINEAR EMBEDDING (MT)

Abstract

In view of the shortcoming that the dimensivity reduction effect of local linear embedding（LLE） is greatly affected by the uniformity of sample distribution, the uniform distance is designed to replace the Euclidean distance in LLE, so the uniform distance local linear embedding（UDLLE） method is proposed, and the effectiveness of UDLLE is verified by typical curves. On this basis, a rolling bearing fault diagnosis method based on UDLLE was proposed, and the diagnosis example was verified. The experimental results show that UDLLE can obtain more effective low-dimensional features than some other methods, reduce the difficulty of subsequent diagnosis and improve the fault diagnosis accuracy.

Keywords

• Local linear embedding;Uniform distance;Rolling bearing;Fault diagnosis