Level: Advancedr
In order to build a real-time abnormal detection monitoring system for roller bearings, the acceleration signal of the roller bearing is preprocessed in 0.5-second units, and then various statistical indices are applied to the preprocessed acceleration signal to define the patterns of NOR (normal), ORF (outer race fault), and IRF (inner race fault) of the roller bearing. Then, a system that monitors abnormal detection in real time is built by learning the patterns using machine learning techniques. Skewness and Kurtosis were selected as statistical indices that recognize the NOR (normal), ORF (outer race fault), and IRF (inner race fault) patterns of the roller bearing acceleration signal by applying the statistical methodology of JMP. In addition, pattern recognition was applied to build an abnormality detection model by applying ANN (Artificial Neural Network), SVM (Support Vector Machines), which are machine learning methodologies provided by JMP, and NL (Nominal Logistics), which is a logistic regression.
