Prediction and optimization of energy consumption in an office building using artificial neural network and a genetic algorithm

Abstract

The aim of this study is to propose a reliable method to optimize the energy consumption of buildings. Also, the most effective input parameters are defined which are used in the energy consumption of a research center building located in Iran as a case study. Accordingly, EnergyPlus software is implemented to evaluate energy consumption and scrutinize the crucial factors numerically. Afterward, a robust artificial neural network (ANN) using multi-layer perceptron model (MLP) is created, trained, and tested to simulate energy consumption in the building. Furthermore, energy optimization is performed by Galapagos plugin based on a Genetic Algorithm considering the critical variables. The main results show that the optimization of the system can mitigate energy consumption by about 35 %. In addition, the outcomes of the sensitivity analysis demonstrate that the number of occupants has the highest influence on the energy consumption of the edifice followed by wall U-value which is related to wall insulation. Finally, the results of computations showed that the trained MLP model proposed in this study can accurately predict energy consumption in the building. To sum up, the proposed model may be applied to similar buildings to predict and optimize energy consumption.