Machine Learning Approach to Predict Cooling Load for Existing Buildings

The objective of this study is to predict air conditioning loads for existing buildings using operational data, weather forecasts and visitor forecasts.The proposed prediction method is based on a neural network approach.However, it is important to note that the proposed method does not learn the entire loads.Loads are divided into factors which can be predicted by traditional thermodynamics and factors which are subject to machine learning.The proposed method has been applied to an example instance using operational data from an underground mall in Kobe, and its validity has been confirmed.