블랙박스 모델을 사용한 가스터빈 동적거동해석 속도 향상

초록

This paper examines the optimal operation strategy of a complex distributed generation (DG) network consisting of various power sources. To predict reliable dynamic behaviors of various power sources and determine the optimal operation strategy of a DG, accurate simulation models are necessary. However, in previous studies on the dynamic simulation of DGs, simple models were used due to limitations in computing time. Those models, however, generally did not accurately reflect the physical characteristics of the power sources. Herein, a novel modeling method for gas turbines is suggested. It predicts the dynamic behaviors of gas turbines both quickly and accurately and can thus be effectively used to simulate an entire DG network. We used gas turbines as our first modeling target because they are considered useful power sources due to their operational flexibility and low emissions. First, a white-box model using physical modeling was developed. Then, a black-box model using a feedforward neural network was constructed on the basis of the simulation results of the white-box model. Finally, a dynamic simulation was performed using the black-box model. The computing speed of the black-box model was 50 times faster than that of the white-box model, and its accuracy was relatively high.

키워드