A Hierarchical Particle Swarm Optimization Framework for Multi-Scale Design of Spectrally Selective Glazing Systems

Abstract

Designing spectrally selective glazing involves a complex, high-dimensional optimization challenge characterized by the strong coupling of millimeter-scale geometric parameters and nanometer-scale photonic thin-film structures. Standard optimization approaches often struggle with the disparity in sensitivities across these scales. To address this, this study presents a hierarchical particle swarm optimization (H-PSO) framework coupled with a transfer-matrix method (TMM) to automate the design of high-performance building envelopes. The proposed framework adopts a two-stage evolutionary strategy: first optimizing the macro-geometry for structural feasibility and ultraviolet suppression, and subsequently refining the nanophotonic SiO₂/TiO₂ interference coatings for nearinfrared rejection under strict manufacturability constraints. Simulation results demonstrate the framework’s capability to locate robust optima, yielding a design with <2% UV transmittance and >80% visible clarity while reducing solar heat gain by 31% relative to baseline models. By effectively bridging optical physics and computational intelligence, this hierarchical approach offers a reproducible, scalable solution for solving multi-scale engineering design problems in sustainable architecture.