Near-zero-energy-consumption buildings (NZEBs) are of great significance for sustainable development, and their design and research have attracted increasing academic attention. To drive and realize the energy saving design and achieve carbon emission and thermal comfort optimization of NZEBs, in this paper, an intelligent optimization method integrating the BIM-DB and PSO-RF-NSGA-III method is established. Multiobjective optimization problems involving NZEBs in four typical climate regions in China are explored. With a typical office building as an example, first, simulation calculations regarding the energy consumption, carbon emissions and indoor thermal comfort in the four climatic regions are performed based on orthogonal tests and BIM-DB. Second, the nonlinear mapping relationships between building design parameters and prediction targets are constructed with the PSO-RF model, which is trained with sample data. The obtained nonlinear mapping relations are used to establish the objective function of NSGA-III, and the multiobjective Pareto-optimal solution set is obtained with the developed PSO-RF-NSGA-III algorithm. Finally, the only optimal solution is determined by using the ideal point method, and reference near-zero-energy-consumption office building parameters are calculated for different climate regions. The conclusions are as follows. (1) The PSO-RF algorithm can efficiently predict building energy consumption, carbon emissions and thermal comfort. In the four regions, the goodness of fit of the three targets is greater than 0.94. (2) Multiobjective optimization can be performed with the proposed RF-NSGA-III intelligent optimization method. After optimizing multiple groups of optimization schemes and adopting energy saving measures, the energy consumption levels in the four climate regions are reduced by 39.72 %, 32.22 %, 26.94 % and 35.37 %, and the other goals are optimized. (3) Index calculations indicate that the optimized building design parameters meet the specified standards for NZEBs, and the main influencing factors and corresponding measures vary from region to region.
Read full abstract