Energy Forecast in Buildings Addressing Computation Consumption in a Green Computing Approach

The decrease of the energy costs and the sustainability in buildings are important to ensure the energy management of buildings. The use of algorithms for forecasting activities is crucial to reduce the energy costs. Green computing plays a vital role on ensuring the energy sustainability of buildings while performing forecasting activities. This paper uses two forecasting algorithms including artificial neural networks and k-nearest neighbors to predict electrical consumptions of a building that monitors consumption data in five minutes periods. It is expected to reach effective prediction accuracies while guaranteeing that both algorithms spend less energy costs on forecasting activities. This case study tests different scenarios with an annual historic and one-week test data to reduce the energy spent in forecasting activities using data simplification strategies which include conversions of five minutes periods to hour schedules and filtering of hour behaviors.