Power Grid Facility Thermal Fault Diagnosis via Object Detection with Synthetic Infrared Imagery

With the rapid development of the economy and society, the daily life of residents and the production activities of enterprises have an increasing demand for electric energy. With the surge of power facilities and power grid load, many potential dangerous factors in the power grid system are always threatening the safety of people's life and property. In order to maintain the safe and stable operation of power grid facilities, condition maintenance and fault diagnosis are particularly important. Traditional manual power facility diagnosis methods often have some problems and deficiencies and will encounter some difficulties in real-world applications. As a kind of instrument diagnostic method, infrared thermal imagery has attracted more and more attention in the electric power industry in recent years with its incomparable advantages. In this paper, we propose to utilize infrared imagery to diagnose the thermal fault of the power grid facilities, especially for the transformers. Specifically, we first employ the deep learning object detection models to locate the power grid facilities in the real world, then the diagnosis system can assess the condition of the facilities according to the infrared thermal imagery. To improve the detection rate of the power grid facilities, we typically synthesize the RGB based image and infrared thermal imagery. The experimental results show that the synthesis technique significantly promotes the detection results.