천연가스 보일러 내부 정상상태 온도장 예측을 위한 디지털 트윈 구축에 대한 연구

Abstract

In this study, a simulation-based digital twin is developed for large-scale industrial natural gas boiler. The digital twin can predict temperature distribution under unexplored operating conditions by a regression model and a limited number of sensor data. The former is generally known as a surrogate model and the latter is known as sparse reconstruction. Proper orthogonal decomposition method is employed together with a kriging regression model for the surrogate model and gappy proper orthogonal decomposition for sparse reconstruction. Three-dimensional simulation data of the natural gas boiler are collected in the parameter space, which consists of excess air rate and swirl vane angle of the device. The surrogate model provides prediction performance within 5% least squared error. With sparse reconstruction, more than 32 measuring points provides prediction performance within 10% least squared error. The results show that the proper orthogonal decomposition has high potential as an analysis method for digital twin and prediction of reacting scalars in industrial combustion devices.