Leakage source estimation by retrospective search for high concentration points in infrared gas concentrarion images

Abstract

In recent years, there has been a concern about the occurrence of gas leakage and fire accidents in plant facilities due to damage caused by aging deterioration of gas piping. However, conventional gas leakage source estimation methods require prior teacher data or calculated data of gas clouds to be compared, which requires large computation time and may not be applicable under unexpected circumstances. Therefore, this study proposes a method to estimate the source of gas leakage by visualizing the leaking gas cloud using an infrared camera and performing an inverse problem analysis based on the concentration information without the need for prior teacher data or calculation data. It is found that the leakage source can be estimated by retrospective search from the gas concentration distribution to the location where the infrared absorption is large. In addition, the leak source estimation is achieved with high accuracy by removing the initial search position data and using only the position data near the leak source in the retrospective search.