Fault Diagnosis of Rotating Machinery using Infrared Stress Method with Deep Learning

Rotating machinery used in industrial applications is designed to withstand the various stresses during operation. When wear or damage causes a failure during operation, the faulty component must be found quickly. Failing to detect a failure can disrupt the smooth formulation of repair plans, leading to increased downtime and higher maintenance costs. Methods for understanding stress distributions in rotating machinery and performing fault diagnosis include strain gauge analysis and vibration analysis. These conventional methods have a long history of research and are well established. However, measuring spatial stress distributions in rotating machinery is difficult due to complexity. To address these limitations, we propose diagnosis method for rotating machinery using infrared cameras. Infrared cameras offer a non-contact and non-destructive condition monitoring method capable of recording temperature distributions over a wide area. Therefore, this method is suited to capture spatial stress distribution. In field situations, however, an emissive distribution of the rotation axis varies within painting conditions. Therefore, there is a need for technology that enables the use of infrared cameras for fault diagnosis of actual rotating machinery. The objective of this study is to capture images of the surface of the rotating shaft of actual rotating machinery using an infrared camera and to predict the results of vibration measurements performed on the same target.