Infrared Image Combined with CNN Based Fault Diagnosis for Rotating Machinery

Infrared (IR) imaging for rotating machinery monitoring and fault diagnosis has gained noticeable attention in recent years since it enables non-contact, on-intrusive and singlesensor based temperature measurements. However, the IR image for fault identification requires the process of artificial feature extraction, which is influenced by personal experience. And the sensitivity of the infrared image signal to various faults in the mechanical equipment is still unclear. Therefore, a rotor platform fault diagnosis method based on infrared image and convolution neural network (CNN) classification algorithm is proposed in this paper, which can automatically select the image feature and identify the fault mode. Firstly, the theoretical basis for the fault diagnosis of rotating equipment using IR image is analyzed. Then, the network structure and parameters were optimized for the IR data. Finally, the proposed method is used to diagnose the rotor platform and compared with the traditional way of feature extraction combined with the SVM, which proved that the proposed method has a higher accuracy. The application of our research was on the detection of conditions such as rotor imbalance, shaft misalignment, bearing looseness, rubbing, and coupling of rubbing and misalignment.