An Optimized Machine Learning Approach for Forecasting Thermal Energy Demand of Buildings

Recent developments in indirect predictive methods have yielded promising solutions for energy consumption modeling. The present study proposes and evaluates a novel integrated methodology estimating the annual thermal demand (DAN), which is considered as an indicator of heating cooling loads buildings. A multilayer perceptron (MLP) neural network optimally trained by symbiotic organism search (SOS), among strongest metaheuristic algorithms. Three benchmark algorithms, namely, political optimizer (PO), harmony algorithm (HSA), backtracking (BSA) are likewise applied compared with SOS. results indicate that (i) utilizing properties building within artificial intelligence framework gives suitable prediction DAN indicator, (ii) nearly 1% error 99% correlation, suggested MLP-SOS capable accurately learning reproducing nonlinear pattern, (iii) this model outperforms other models such MLP-PO, MLP-HSA MLP-BSA. discovered solution finally expressed explicit mathematical format practical uses future.