Fine-grained modeling and optimal control methods via video-based positioning for multi-occupant smart lighting systems

Abstract

This paper proposes fine-grained modeling and optimal control methods for the Occupant-Centric Smart Lighting System (OCSLS) in open-plan office environments. The system can utilize the occupant's video data to optimize luminaires dimming, achieving significant energy savings and improved illumination comfort. The main contributions are as follows. Firstly, by utilizing 3D reconstruction of human keypoints extracted from multi-view video, a robust non-invasive positioning algorithm for multi-occupant is proposed, achieving a positioning error of less than 0.3 m. Secondly, a novel Illumination Demand Matrix (IDM) is generated through the integration of the fine-grained personal illumination preference model and the location data of occupants, and an Illumination Supply Matrix Library (ISML) is obtained considering the illuminance characteristics and locations of luminaires. Thirdly, based on the optimization principle of matching illumination demand and supply, an Occupant-Centric Control (OCC) strategy is developed using the similarity search of the proposed IDM and ISML with a balance between energy saving and comfort. The experimental results based on the established verification platform indicate a reduction in energy consumption over 20 % and an occupant satisfaction rate above 85 %.