Passive Thermal Imaging-based Fault Detection in Induction Motor Under Varying Speed Conditions

Abstract

Induction motors (IM) are commonly utilized as the prime movers in various industrial applications because of their simplicity, reliability, and minimal maintenance cost. Fault diagnosis of IM is one of the primary issues that seeks to make sound maintenance decisions to preserve a system's integrity and safety while reducing unplanned downtime and lowering maintenance costs. Most fault diagnosis approaches for IMs are based on analyzing vibration signals captured at constant rotating conditions. Those vibration signal-based methods are less capable at under-speed varying conditions. This paper proposed a Passive Thermal Imaging (PTI) based fault diagnosis approach for IM at varying speed conditions to deal with these issues. Firstly, various thermal image frames are extracted from the captured thermal video from the healthy and faulty IM at varying speed conditions. Thereafter, Residual Network (ResNet) is used for extraction of features, followed by further classification using Support Vector Machine (SVM) at various fault conditions. The findings demonstrate that the suggested technique outperforms traditional vibration-based methods in identifying various IM faults at varying speed conditions.