기계학습을 이용한 냉동기 최적 가동/정지 시점 결정 모델 개발

In this study, the authors developed ANN(Artificial Neural Network) models using Building Energy Management System (BEMS) data of a real-life office building. The ANN models are used to predict chilled water temperature leaving from a chiller system at the sampling time of 15 minutes for the time horizon of two hours. Based on the prediction of the ANN models, optimal control (optimal start and time) strategies of a chiller and Air Handling Unit (AHU) were developed. Due to ANN models, the operation time of a chiller can be saved by 4.7%.