Gearbox Fault Detection under Varying Load Based on Autoregressive Modeling of Vibration Data

摘要

The development of the fault detection for gearbox systems has received considerable attention in recent years. Since most gearbox systems operate under varying load which affect the vibration signature of the system and generally make it difficult to recognize the occurrence of a fault, we develop the fault detection schemes based on the residuals calculated using autoregressive models fitted to the data obtained under various load conditions of a gearbox. This fault detection schemes is then proposed and tested using full lifetime data obtained from a gearbox operating from a new condition to a breakdown.