Comparative Study of Multi-Objective Optimization Processes for Building Performance

Growing demands for energy efficiency, thermal comfort, daylighting, cost-effectiveness, and carbon mitigation have made simulation-based multi-objective building performance optimization an important research area. However, comparisons of algorithm applicability and performance remain scarce. A review and comparison of multi-objective building performance optimization is presented by combining literature review and bibliometric analysis. This review covers five key aspects of building performance optimization: optimizing problem construction (building type, optimization objectives, and design variables), data preprocessing methods (including data preprocessing, sensitivity analysis, and clustering analysis), surrogate model selection, optimization algorithm selection, and multi criteria decision-making. The results indicate that building performance optimization has evolved from energy-focused optimization to multi-objective trade-offs involving energy, comfort, daylighting, cost, and carbon emissions. Among surrogate models, artificial neural network shows stable performance in multi-output regression, whereas linear regression is more appropriate for single-output problems. Surrogate-assisted frameworks integrating artificial neural network, Gaussian process regression, random forest, or radial basis function models generally improve computational efficiency under limited evaluation budgets. NSGA-II remains the most widely used and reliable algorithm for bi-objective and tri-objective optimization. MOEA/D shows limited suitability for many-objective building optimization, while particle swarm optimization converges rapidly for continuous variables but is less robust for constrained mixed-variable problems. Further, we elaborated on the cutting-edge research directions of multi-objective optimization for building performance and comparative studies of algorithms based on different building optimization problems, while identifying opportunities and challenges for future applications of building performance optimization.