Abstract
This paper proposes a method of detecting fault conditions of escalator using several statistical parameters of vibration and sound. We have developed an inspection step which is equipped with an accelerometer and a microphone, and it can easily acquire the sensor data of the running escalator in a short time. The acquired data is parameterized by amplitude of vibration, kurtosis of vibration, cyclic vibration level, kurtosis of sound, and kurtosis of high-pass sound. When at least one statistical parameter exceeds a threshold, the escalator is judged as the fault condition. The system tests were conducted to evaluate the performance for several conditions. The fault thresholds of the statistical parameters were determined based on many escalator data on the normal condition. The test results showed that the statistical parameters exceed the threshold at the most of the fault cases. Therefore, it is possible to perform fault diagnosis using the differences of the statistical parameters between normal and fault conditions.
