Gearbox Fault Diagnostics using AE Sensors with Low Sampling Rate

Acoustic Emission (AE) sensors have been investigated as a potential tool for machinery health monitoring and fault diagnostics. While AE sensors could possibly provide higher fault detection sensitivity compared with vibration sensors, they also have some drawbacks. AE sensors generally output signals in the range of several hundred kHz up to several MHz, making the AE data sampling and processing costly. In this paper, a method on gearbox fault diagnosis using AE sensors with a low sampling rate is presented. In the presented method, a heterodynebased frequency reduction technique is employed to demodulate the AE signals and shift signal frequencies to a low range. The AE signals are sampled at a rate as low as 20 kHz, which is typically used for vibration signals sampling in industrial applications. Time synchronous average signals are computed from AE signals sampled at the low rate and used to compute condition indicators for gear fault diagnosis. The diagnostic performance of the condition indictors computed using both AE and vibration data sampled at the same rate of 20 kHz is compared. Both AE and vibration data is collected on a notational split torque gearbox with different levels of seeded tooth faults. The results have shown that AE signals sampled at a low rate suffice for fault detection purpose and are promising for damage level diagnosis. Compared with vibration analysis results, AE provides better fault diagnosis sensitivity to tooth damage level. Since AE is normally unaffected by the machine resonance, it can potentially offer more stable and reliable performance under the same sampling condition as vibration.