건물 에너지 분야의 인공지능 기반 연구 동향 분석 - 해외 저널 논문 중심으로 -

Purpose: Recently, there are many research projects conducted to achieve smart cities. Smart cities consist of smart buildings that include efficient energy supply and consumption systems. The Artificial Intelligence (AI) technologies became useful tools for this purpose due to their reliability of prediction accuracy and credibility. It is very important to better understand how the AI algorithms work and can be applied for specific areas of energy efficiency in buildings. This paper presents how AI technologies, such as Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and Long Short-Term Memory (LSTM), are currently being utilized in the energy efficiency research in buildings. Method: International journal papers are reviewed especially for those utilizing ANN, CNN, RNN, and LSTM algorithms in building science and technologies. In-depth analyses are conducted comparing specific approaches, research outcomes, advantages, and disadvantages of key papers. Result: Findings show that the ANN, CNN, RNN, and LSTM algorithms are mainly used for the prediction of building energy loads and system energy uses. Compared to other AI algorithms, the LSTM algorithms have higher prediction accuracies due to the characteristics of LSTM structure.