Optimized Design Research on Daylighting Performance of Cold Land Buildings Based on Improved Neural Network
Liu Lei,Sun Cheng,Liu Ying,Leng Hong,Yang
DOI: https://doi.org/10.2478/amns-2024-0730
2024-01-01
Applied Mathematics and Nonlinear Sciences
Abstract:This study delves into optimizing daylighting in buildings in cold regions, employing an innovative neural network approach to enhance natural lighting efficiency. Cold climates present unique challenges for daylighting, making it essential to improve indoor lighting conditions, reduce energy usage, and enhance occupant comfort. Traditional design methods fall short in optimizing daylighting due to their inability to effectively navigate complex environmental factors and building configurations. We introduce an advanced neural network model that pioneers efficiency and innovation in the daylighting design of cold buildings. This model leverages the GA-PSO-BP framework, integrating Genetic Algorithms (GA), Particle Swarm Optimization (PSO), and Back-Propagation (BP) neural networks to create a potent optimization tool. Our approach focuses on refining key design parameters such as building orientation, floor height, plan depth, and external window design. Notably, specific adjustments to building orientation and floor height significantly boost daylight autonomy (DA) and helpful daylight illuminance (UDI) while maintaining the daylight glare probability (DGP) within optimal limits. Our findings reveal that optimizing building orientation can elevate DA and DGP values by 4.756% and 0.037325, respectively. Similarly, adjustments to floor height can enhance DA, UDI, and DGP values to 51.833%, 51.278%, and 0.361377, respectively. This refined neural network model demonstrates a robust capability to improve daylighting performance in cold-region buildings, offering fresh perspectives and methodologies toward the sustainable evolution of architectural design.