Cycloidal Gear Faults’ Detection of Industrial Robot Joint Based on IAS Signal Under Operating

Abstract

Conventional approaches to fault detection in industrial robot joints predominantly utilize vibration sensors. However, spatial constraints and intricate wiring requirements often hinder the application and significantly escalate detection costs. Furthermore, the dynamic and nonstationary operating conditions of robot joints are characterized by directional reversals, variable speeds, and incomplete cycles. To overcome these issues, this study proposes an encoder signal-based fault detection method. The methodology encompasses several key steps: first, the separation of instantaneous angular speed (IAS) signals based on rotation direction; second, the removal of trend components and amplitude modulation effects induced by variable speeds; and third, the reconstruction of complete fault cycles from fragmented signal segments. To enhance diagnostic precision, the narrowband demodulation technique is employed to detect specific fault types, including tooth wear and tooth-root cracks, in cycloidal gears. The validity of the proposed method is substantiated through simulations and experimental evaluations.