Abstract
This paper is to find simple and effective fault detection methods related to monitoring system of reduction unit for high-speed railway. Two tests, the specimen and the full scale reduction unit of high-speed train test, are conducted. The specimen test used spur gears with 23 teeth such as the no-fault, cracked and broken spur gears, and the full scale reduction unit test employed no-fault and broken helical gears with 51 teeth. The vibration data of tests was analyzed by conventional methods such as RMS, crest factor, kurtosis and Sideband Energy Ratio (SER). The results analyzed by various methods are as follows. RMS, Crest factor and kurtosis could distinguish no-fault and breakage except crack in the specimen test. SER of 1st Gear Mesh Frequency (GMF) increased gradually when gears changed from no-fault to breakage condition. In the full scale reduction unit test, RMS was similar to the specimen test. Crest factor and kurtosis differed from the specimen test results, that is, the levels of crest factor and kurtosis of breakage condition were not larger than no-fault condition. SER could distinguish no-fault and breakage. Through these results, RMS and SER seem to be more effective than crest factor and kurtosis when applied to monitoring of the reduction unit of high-speed train.
