Detection of Gear Faults based on Acoustic Emission Signals Using Logistic Model Tree Classification Method

Gearbox is a main component of rotating machines. Any fault in gearboxes not detected and removed on time, results in stopping the system or damaging it. So, early detection of faults is of critical importance. Here, in order to automatic detection of gear faults based on Acoustic emission signals Logistic Model Tree (LMT) classification method isapplied. Wavelet transform is used to analyse Acoustic Emission signals recorded from a defected gear. Then, some discriminative features are extracted from different frequency bands. From many different extracted features, the most relevant ones are selected using feature selection algorithms. At last, a Logistic Model Tree is used to classify theundefected and defected gears; also, the types of defects are recognized. Undefected and defected gears are discriminated with 99% of accuracy and types of defects are correctly identified with the rate of 80%.