Thermal image based fault diagnosis for rotating machinery

摘要: Abstract Infrared imaging is crucial for condition monitoring as the thermographic patterns will differ depending on fault or machine condition. Currently, a limited number of faults have been studied using thermal imaging. Therefore, this paper proposes novel automatic detection system infrared imaging, focussing bearings rotating machinery. The set bearing monitored contain which state-of-the-art techniques no general solutions such bearing-lubricant starvation. For each fault, several recordings are made different to ensure generalization fault-detection system. contains two image-processing pipelines, with their own respective purposes. first pipeline focusses detecting rotor imbalance, regardless faults. second faults, whether balanced not. Within pipeline, imbalance detected by differencing consecutive image frames subsequently summarized distribution along axes. three features introduced standard deviation temperature, Gini coefficient, and Moment Light. final able distinguish between all eight conditions an accuracy 88.25%.