リカレントニューラルネットワークによるビル空調システムの消費電力推定

抄録

This article proposes a prediction method of power consumption for building air conditioning systems with multiple indoor units per outdoor unit. As a first step, a recurrent neural network (RNN) is used to establish a model that estimates power consumption from current states of the building. The estimation accuracy is evaluated by leave-one-out cross validation using data that was recorded at an air-conditioned building for 14 days. The evaluation results showed that the power consumption for a day can be approximately estimated by a RNN trained using room temperatures and outside temperature of the other 13 days. The next step would be to predict the power consumption using an ideal room temperature profile and forecasted outside temperature.