Akademik Veri Yönetim Sistemi
Light and Engineering, cilt.34, sa.2, ss.35-48, 2026 (SCI-Expanded, Scopus)
Improving the energy efficiency of indoor lighting systems is a key strategy for reducing electricity consumption without compromising visual comfort. This study presents a performance assessment of Artificial Intelligence (AI) based daylight prediction models for open-loop indoor lighting control and compares their energy-saving impact with a reference closed-loop system. Real-world daylight data collected over a two-year period from a public building forms the basis of the analysis. The closed-loop system employs sensors at each measurement point, while the open-loop alternative leverages machine learning-based daylight prediction models using parameters such as room dimensions, wall reflectance, window size, measurement point distance from windows, and spatial orientation. Eight regression-based prediction models were developed and evaluated, among which the Support Vector Regression (SVR) model demonstrated the highest accuracy, while Theil-Sen Regressor yielded the lowest. The energy analysis was conducted specifically for electrical lighting electricity consumption, and results reveal that closed-loop control systems can reduce annual electricity consumption for lighting by up to 52 %, whereas open-loop systems achieve savings between 33 % and 40 %. Despite its superior performance, the closed-loop approach incurs higher installation and operational costs, resulting in a payback period of approximate ly from 8 to 10 years. These findings highlight the potential of AI-assisted open-loop lighting control systems as a cost-effective solution for retrofitting existing buildings.