Smart Estimation Model for Energy Performance Certificates of Residential Buildings

Abstract

The wasted amount of energy has become a critical problem for many countries due to limited energy sources and energy production costs. These forced countries to increase energy usage awareness by making regulations in the construction sector, which might dramatically decrease energy consumption cause of the size of the domain. One of them is standardizing heating and cooling loads (HL/CL) to avoid energy waste. HL and CL need an advanced engineering process because of different parameters such as the thermal characteristics of the building, hot water supply, passive solar systems, etc. Hence, it can only be carried out by expert engineers in calculations. In this paper, a classification model as the decision support system is proposed for predicting the energy consumption of residents which is an efficient indicator of architectural features of the construction about energy consumption concept. The data is collected from architectural projects and energy performance certificates. Multilayer Perceptrons, Bagging, and Random subspace are used to predict the energy class of buildings. Based on findings, the most accurate results were achieved by Bagging. Moreover, main input features affecting the prediction performance of HL were revealed and classification success was observed.