Vibration Signal for Bearing Fault Detection using Random Forest

Abstract Based on the chosen properties of an induction motor, a random forest (RF) classifier, machine learning technique, is examined in this study for bearing failure detection. A time-varying actual dataset with four distinct states was used to evaluate suggested methodology. The primary objective research defect detection accuracy RF classifier. First, run loops that cycle over each feature data frame corresponding daytime index determine states. There were 465 repetitions inner race fault and roller element test 1, 218 outer 2, 6324 3. Secondly, task find typical procedure differentiate between normal erroneous data. Out 3 tests, (22-23) % obtained since every beginning degrade usually exhibits some form spike many locations, or not operating at its optimum speed. Thirdly, display comprehend 2D 3D environment, Principal Component Analysis (PCA) performed. Fourth, algorithm classifier recognized frame’s predictions, which 99% correct bearings, 97% accurate races, 94% faults. It thus concluded proposed capable identify