48 / 2021-10-09 16:16:44

Diagnosis Method of Metal Surface Temperature Rise by Visible Images and Machine Learning

AI-driven technology > Intelligent Anomaly Detection, Fault Diagnosis and Prognostics

This paper first introduces and summarizes our previous work on the temperature measurement method of metal surfaces by visible images and machine learning(ML). This method provides a low-cost, convenient and accurate detection method for the abnormal temperature rise in electrical equipment. Based on our previous work, this paper conducts the temperature rise (temperature difference) experiments of real metal devices used in electrical equipment in real sunlight, including aluminum alloy clamps, copper wire bars, and copper tinned wire bars. The results confirm the applicability of this method in engineering practice. In addition, the experimental results in this paper show that the measurement system has a smaller prediction error when the sunlight illuminance is large. Therefore, sufficient illumination helps to ensure the accuracy of the measurement. This will provide guidance for the application of this method in practice.