Quaternion Based Omnidirectional Machine Condition Monitoring System

Thermal monitoring is useful for revealing some serious electrical problems in a factory that oftengo undetected until a serious breakdown occurs. In factories, there are various types offunctioning machines to be monitored. When there is any malfunctioning of a machine, extra heatwill be generated which can be picked up by thermal camera for image processing andidentification purpose. In this paper, a new and effective omnidirectional machine conditionmonitoring system applying log-polar mapper, quaternion based thermal image correlator andmax-product fuzzy neural network classifier is proposed for monitoring machine condition in anomnidirectional view. With this monitoring system, it is convenient to detect and monitor theconditions of (overheat or not) of more than one machines in an omnidirectional view captured byusing a single thermal camera. Log-polar mapping technique is used to unwarp omnidirectionalthermal image into panoramic form. Two classification characteristics namely: peak to sideloberatio (PSR) and real to complex ratio of the discrete quaternion correlation output (p-value) areapplied in the proposed machine condition monitoring system. Large PSR and p-value observe ina good match among correlation of the input thermal image with a particular reference image,while small PSR and p-value observe in a bad/not match among correlation of the input thermalimage with a particular reference image. Simulation results also show that the proposed system isan efficient omnidirectional machine monitoring system with accuracy more than 97%