Optimization of Winter Lighting Performance of Triple-Glazed Windows Based on Neural Networks

Abstract

With the escalating global energy crisis and the introduction of dual carbon targets, building energy conservation has become a key issue for China’s sustainable development. Triple-glazed windows, due to their excellent thermal insulation and light-transmitting potential, are widely used in modern energy-efficient buildings. This study aims to explore the optimal solution for the combination of three glass thicknesses using advanced intelligent optimization algorithms, in order to maximize solar energy utilization during cold winters. The study first established an accurate optical model based on the Transfer Matrix Method (TMM), and then used this physical model to generate a large-scale dataset to train a feedforward neural network surrogate model that can efficiently predict the transmission performance of glass. Finally, by combining the differential evolution algorithm, the optimal thickness combination that maximizes transmittance was successfully found within the vast design space. This study provides a rapidly achievable intelligent optimization path for the optical design of lowenergy building envelopes.