Early Fault Detection of Robotic Joints Based on Time–Frequency Analysis of Motor Current Signature

Robotic joints, which are critical fixtures consisting of electrical motors, high-speed planetary gears, and low-speed cycloid wheels, are prone to friction and wear. Early fault detection and diagnosis of joint conditions are fundamental to providing relevant information for implementing maintenance. In this study, a motor current signature analysis (MCSA)-based method is proposed for monitoring and diagnosing the early faults in industrial robots. The research leverages the characteristics of motor current changes related to load variations, introducing time–frequency analysis (TFA) for early fault detection. The Fourier synchro squeezed transform (FSST) method provides energy-concentrated time–frequency representations, effectively capturing changes in robotic joint motion states through extracted time–frequency ridge features. Experimental studies were carried out based on a six-degree-of-freedom industrial robot with an eccentric mass on robot arms to simulate the early fault of joint frictional and wear. TFA results and corresponding features allow accurate fault magnitude and location resolution, demonstrating the effectiveness of the proposed method.