ARIMAモデルによる空調熱負荷予測

Abstract

Conventional commercial building air conditioning systems typically employ heat storage systems that have sufficient heat source capacity to achieve the utilization of recovered waste heat and night electric power, and to meet peak instantaneous loads. In light of the increasing popularity of these systems, a need exists for the prediction of the hourly load profiles of air conditioning system took place in a systematic fashion.This work is concerned with a prediction problem for the hourly loads by taking account the Autoregressive Integrated Moving Average (ARIMA) model using the past load data. This is a modified form of the ARMA model. Some of technical considerations are examined.At the end of each day (at 22:00), the modelling is first done for a set of observed data by taking into account the ARIMA model. In this modelling procedure, the historical load profile data are directly used but the ambient temperature profile data are not used. Next, this ARIMA model obtained above is used to predict the hourly load for the next day. The load profiles are updated every hour on the basis of the newly obtained load data for a given clock hour.The differences between predicted and actual load are examined for days 100 (2544 observations) from February to July in 1987. There is generally good agreement between predicted and actual loads. The strategy algorithms are now executed to update the nominal operation of the heat storage system to meet the updated load profile for the subsequent hours.