Research on Fine-Grained Modeling and Optimal Control Used for Occupant-Centric Smart Lighting Systems

作     者：Ning, Chenguang Shen, Dongdong Dong, Yanzhi Wang, Yingjie Duan, Peiyong 

作者机构：School of Information Science and Engineering Shandong Normal University Jinan250014 China School of Physics and Electronic Information Yantai University Yantai264005 China School of Computer and Control Engineering Yantai University Yantai264005 China Faculty of Electronics Electrical and Control Qilu University of Technology Shandong Academy of Sciences Jinan250353 China 

出 版 物：《SSRN》 

年 卷 期：2024年

核心收录：

主　　题：Nearest neighbor search 

摘      要：In order to reduce energy consumption of the lighting system and improve the visual comfort of occupants in the illuminated area, it is essential to effectively and precisely control the illumination of the luminaires. The smart lighting systems can utilize the occupant s positional data to adjust lamps switching, leading to a significant decrease in energy consumption and an enhancement in illumination comfort. However, there are still several challenges associated with achieving precise and non-invasive multi-person positioning for smart lighting systems, and the implementation of the Occupant-Centric Smart Lighting System (OCSLS) is still pending. In this paper, using 3D reconstruction of human keypoints extracted from multi-view video, the robust positioning algorithm for multi-person is proposed. The algorithm achieves a positioning error of less than 0.3 meters by combining the constraints of epipolar geometry and the appearance characteristics of occupants for multi-view matching. Subsequently, a novel fine-grained model of the OCSLS is proposed through the integration of the Illumination Demand Matrix (IDM), which is derived from the distribution of occupant position, and the Illumination Supply Matrix Library (ISML), which is based on the characteristics of lighting area luminaires. Using the OCSLS model proposed in this study, the Occupant-Centric Control (OCC) strategy is constructed through the optimization algorithm based on the autoencoder and the K-Nearest Neighbor (KNN) algorithm. A verification platform has been created, and the experimental results indicate a decrease in the energy usage of the lighting system by over 40%. © 2024, The Authors. All rights reserved.