Electricity load forecasting using clustering and ARIMA model for energy management in buildings

Abstract Understanding the energy consumption patterns of buildings and investing efforts toward energy load reduction is important for optimizing resources and conserving energy in buildings. In this research, we proposed a forecasting method for the electricity load of university buildings using a hybrid model comprising a clustering technique and the autoregressive integrated moving average (ARIMA) model. The novel approach includes clustering data of an entire year, including the forecasting day using K‐means clustering, and using the result to forecast the electricity peak load of university buildings. The combination of clustering and the ARIMA model has proved to increase the performance of forecasting rather than that using the ARIMA model alone. Forecasting electricity peak load with appreciable accuracy several hours before peak hours can provide the management authorities with sufficient time to design strategies for peak load reduction. This method can also be implemented in the demand response for reducing electricity bills by avoiding electricity usage during the high electricity rate hours.