Gears are important mechanical components for power transmissions. Tooth wear is one of the most common failure modes, which can present throughout a gear's lifetime. It is significant to accurately monitor gear wear progression in order to take timely predictive maintenances. Motor current signature analysis (MCSA) is an effective and non-intrusive approach which is able to monitor faults from both electrical and mechanical systems. However, little research has been reported in monitoring the gear wear and estimating its severity based on MCSA. This paper presents a novel gear wear monitoring method through a modulation signal bispectrum based motor current signal analysis (MSB-MCSA). For a steady gear transmission, it is inevitable to exist load and speed oscillations due to various errors including wears. These oscillations can induce small modulations in the current signals of the driving motor. MSB is particularly effective in characterising such small modulation signals. Based on these understandings, the monitoring process was implemented based on the current signals from a run-to-failure test of an industrial two stages helical gearbox under a moderate accelerated fatigue process. At the initial operation of the test, MSB analysis results showed that the peak values at the bifrequencies of gear rotations and the power supply can be effective monitoring features for identifying faulty gears and wear severity as they exhibit agreeable changes with gear loads. A monotonically increasing trend established by these features allows a clear indication of the gear wear progression. The dismantle inspection at 477 hours of operation, made when one of the monitored features is about 123% higher than its baseline, has found that there are severe scuffing wear marks on a number of tooth surfaces on the driving gear, showing that the gear endures a gradual wear process during its long test operation. Therefore, it is affirmed that the MSB-MSCA approach proposed is reliable and accurate for monitoring gear wear deterioration.

Abstract：Motor current signal analysis has been an effective way for many years to monitor electrical machines.However, little research work has been reported in using this technique for monitoring their downstream equipment due to difficulties in extracting weak fault signals in a noisy measured current signal. This paper investigates the characteristics of electrical current signals measured from an induction motor for monitoring faults from a downstream gear transmission system by a novel modulation signal (MS) bispectrum method. Both analytic analysis and experimental study are conducted based on a motor drive system and have found that the increase of bispectral peaks can be the basis of mechanical fault diagnosis. Particularly, a fault on a gear will causes a larger increase of the bispectral peak at both the gear shaft frequency and accompany with a smaller increase in the relating shaft frequency. However, a shaft misalignment only leads to a bispectrum peak increase at the shaft bifrequency along.

No abstract