The Community for Technology Leaders
Pacific-Asia Workshop on Computational Intelligence and Industrial Application, IEEE (2008)
Dec. 19, 2008 to Dec. 20, 2008
ISBN: 978-0-7695-3490-9
pp: 869-873
This paper proposed a method of geometrical learning (GL) for identifying mechanical fault patterns and its grades. The method aimed at optimal covering for each class, and constructed some complicated geometry bodies (CGBs) to cover one class samples distributing in the feature space approximately. The rolling bearing was studied by GL-based method in this paper, firstly, energy spectrum was extracted for features by wavelet packet transform, and then the feature vectors data was done topological analysis to construct CGBs for each class by the training algorithm of geometrical learning; secondly, the independent test set was tested by the identification algorithm of geometrical learning. On the condition of experiment, the fault patterns and its grades (from gently to severity) could be identified correctly. Moreover, the identification ability of the GL-based method was compared with that of Supper Vector Machine (SVM), the experimental results showed that GL-based method performed better than SVM.
Lijun Ding, Hongwu Li, Liang Hua, "Mechanical Fault Patterns Identification and Grades Cognizance Based on Geometrical Learning", Pacific-Asia Workshop on Computational Intelligence and Industrial Application, IEEE, vol. 01, no. , pp. 869-873, 2008, doi:10.1109/PACIIA.2008.44
89 ms
(Ver 3.3 (11022016))