확률적 모델 학습을 통한 냉장고 내부 온도 모델 구현

Abstract

The purpose of this study was to get high level of performance in system identification of refrigerator. This research conducted several data-driven methods like as multi-layer perceptron (MLP) and long short-term memory (LSTM) of deep learning methods and equation based bayesian optimization and hyper-band. The real data was gathered by several experiments and established system identification resulting in accuracy of 0.27, 0.18 and 0.02 as LSTM, MLP and equation based bayesian optimization.