Real-time wear monitoring in plastic gears using drive torque signal diagnostics

Abstract

Excessive wear represents a prevalent failure mechanism in plastic gears, necessitating online monitoring for intelligent operation and maintenance. To address challenges in direct wear measurement arising from complex geometry and harsh operating conditions, this study proposes a torque signal-based method for staged wear assessment and failure diagnosis. This approach leverages the established correlation between drive torque fluctuations and wear progression observed during testing. Utilizing Symmetrized Dot Pattern (SDP) transformation, one-dimensional torque signals are converted into two-dimensional images for enhanced feature characterization. Torque signals spanning the complete service-life cycle at distinct wear stages were acquired to develop a YOLO11-CLS-based wear classification model. By monitoring torque variations during gear meshing in real-time, the method achieves wear staging and failure diagnosis. Experimental results demonstrate significant morphological differences in Polyamide (PA) gear debris across operating conditions: lower parameters produce finer, more uniform debris. PA gear wear evolution follows a distinct three-stage process (running-in → stable wear → severe wear), with torque signals enabling stage differentiation at >96 % classification accuracy. The method provides effective failure diagnosis for PA gears while offering practical implementation advantages. This research contributes fundamental data for PA gear load-bearing design and advances condition monitoring methodologies for plastic gear transmissions.