Research on Insulating Oil Gas Analysis and Fault Prediction Based on the Edge Computing Platform of the Internet of Things

Abstract

The analysis of dissolved gas in insulating oil is essential for judging the abnormality or potential failure of oil-filled electrical equipment such as transformers or high-resistance. Currently, oil chemical analysis experiments mainly rely on manually taking oil samples and sending them to testing laboratories for manual analysis and judgment. This process takes a long time, and the oil samples are prone to oxidization and deterioration during transportation, which significantly reduces the detection efficiency and judgment accuracy. Based on the edge computing platform of the Internet of Things, this paper builds the framework of the online insulating oil gas analysis system. It uses the Least Squares Twin Support Vector Regression machine (LSTSVR) as the calculation model on the edge side and then carries out the accuracy and accuracy test. The test results show that the system can automatically sample and analyze in real-time, shorten the detection delay, and provide accurate data results and fault range prediction.