Induction motor electric signals contain torsional vibration features of the drivetrain through electro-mechanical coupling. By taking the induction motor itself as a build-in sensor of the drivetrain, gear fault diagnosis can be realized without mounting extra sensors on or even near the geartrain. Such merit is appealing for geartrain working under extreme environment. Available references on this topic mostly focus on stator current signal analysis. However, the relatively-weak gear fault signatures in stator current urge the discoveries of new schemes. In this work, aiming at enhancing the capability of electric-signal-based gear fault diagnosis, four new forms of induction motor electric signal are investigated. A uniform analytical model for current/voltage signals of stator/rotor windings or rotor shaft is established, to elaborate and compare their effectiveness in revealing gear fault signatures. Finite-element numerical simulations and laboratory experiments validate the analytical derivations, and quantitively compare the strength, saliency, monotonicity, and linearity of the contained fault signatures. The results pinpointed the advantages of rotor current/voltage and shaft voltage analyses over the conventional stator current analysis in incipient gear fault diagnosis and fault severity assessment. These explorations aim to motivate and guide the utilization of new forms of electric signals in but not limited to the field of electric drive maintenance.
2 January 2024 in IEEE Transactions on Power Electronics
In rotating machinery fault diagnosis, the accurate rotational speed is essential for interpretating the repeating fault-induced characteristics. However, installation of extra tachometer can be costly. The widely used induction motor not only drives the powertrain, but also serves as an embedded tachometer via electromagnetic coupling. Available trials on motor-current-based rotational speed estimation either demand the detailed motor configurations or oversimplify the spectral complexity. In this work, a novel rotational speed estimation method is proposed based on adaptive identification of the speed-related modulation sidebands in the induction motor stator current signal. No motor parameter like the numbers of pole pairs or rotor slots is required, and the rotor rotational speed can be automatically extracted from the numerous spectral peaks. The effectiveness of the proposed method is validated by three stator current datasets measured from different induction motors. The obtained rotational speed supports the diagnoses of gear chipping and shaft misalignment in the drivetrain. The feasibility, accuracy, and efficiency of the proposed rotational speed estimation method are quantitatively evaluated with examples.