Desenvolvimento de sistema para análise de termograma visando detectar falhas utlizando processamento digital de imagens

Abstract

Thermography allows the visualization of the infrared spectrum for analysis of thermal images, which are thermograms that allow direct observation of the distribution of body heat. This inspection technique, through the decomposition of the infrared spectrum color temperature radiated by the object, measures the distribution of surface temperature. The purpose of performing this inspection is to detect and diagnose temperature rises in electrical components and thus reduce the interruption of tasks. The use of thermography is quite common in several industrial sectors. We can use it in predictive maintenance through inspection of electrical panels, substation, central air conditioning, among others. Among the key advantages of this technique is the reduction in corrective maintenance, reduction of maintenance cost and low power consumption. Other advantages are to carry out measurements without physical contact with the object, checking equipment in full operation and inspection of large areas in short time. In this sense, this paper aims to develop an open-source program for Windows platform able to analyze thermograms captured by infrared cameras and make these diagnoses based on the variation of the color tones of the RGB system. Tests based on the NBR 5410 were performed in order to prove the veracity of the diagnosis made by the developed system using thermograms captured by an infrared camera. The computational tool developed proved to be accurate and has proved an efficient way to inspect industrial equipment.