There are many kinds of rotating components mounted on railway vehicles such as traction motors, generators, and traction gears. The failures of such components sometimes lead to service disruptions and/or accidents. Therefore, it is important to detect their abnormalities at an early stage and prevent their failures. In general, vibration monitoring is an effective abnormality detection method for rotating components. However, detection of the vibration of those components is complicated by vehicle vibration and varied operational status. To address these issues, the authors have proposed an abnormality detection method using vibration octave spectra and machine learning. As a means of verifying the proposed method, engine tests are conducted using auxiliary drive shafts with two simulated abnormalities. The test results indicate that the proposed method enables us to detect them and distinguish between them.
View full abstract