Studying the Implementation of State-of the Art Meta-heuristics to Boost Energy Conservation of Residential Buildings

Residential households consume considerable portions of energy use and CO2 emissions. Accordingly, fast and accurate prediction of magnitudes of heating and cooling demands (HD and CD) are indispensable to facilitate delivering optimum designs of energy-efficient buildings. This research paper investigates and compares nine state-of-the-art bio-inspired meta-heuristics capitalizing on their usefulness in anticipating amounts of HD and CD in residential buildings. These meta-heuristics are coupled with Elman recurrent neural network (ERNN) to build reliable energy prediction models. They involve 1) whale optimization algorithm (WOA), 2) chimp optimization algorithm (CHOA), 3) dragonfly algorithm (DA), 4) multiverse optimization algorithm (MVO), 5) mountaineering team-based optimization algorithm (MTBO), 6) antlion optimization algorithm (MVO), 7) sine–cosine optimization algorithm (SCA), 8) gold rush optimization algorithm (GRO), and 9) dung beetle optimization algorithm (DBO). The accuracies of these models are appraised using various quantitative and visual comparisons. It is concluded that the MVO-based model is the most accurate predictive model of heating demands with MAPE (9.8%), RAE (0.249), IAE (0.102), MAE (2.275) and RRSE (0.333). It is also elucidated that the MTBO-based is the most powerful forecasting model of cooling demands with MAPE (8.56%), RAE (0.26), IAE (0.091), MAE (2.233) and RRSE (0.437). In addition, the DA-based model is the least preferred in the anticipation of HD (MAPE= 15.9%, RAE=0.401, IAE=0.164, MAE=3.665 and RRSE=0.51) and CD (MAPE= 23.17%, RAE=0.691, IAE=0.241, MAE=5.926 and RRSE=1.012). With that said, it is underlined that the reported models can usher sustainable architectural designs that can support energy conservation of residential buildings.