摘要
近年来,商业建筑的智能运行控制策略被广泛研究,而照明系统是重要的系统之一。大多数商业建筑中,照明系统的控制方法基于被动式红外传感器(PIR)收集的实时在室数据。现有的照明系统控制利用时间延迟控制方法。然而,即使房间有人,PIR传感器的监测误差常常导致照明系统出现"错关"的现象。因此,本研究基于房间历史人员在室状态提出了一种基于时序分析的模型预测控制方法(MPC)以解决此问题,利用历史在室概率和时间序列特征对实际人员在室情况进行预测。将新提出的模型控制方法与传统的照明系统控制方法进行比较,结果表明在案例测试期间,新提出的控制方法实现了97.4%的人员在室预测准确率,照明系统的误关次数从传统时间延迟法的79.5次/d,降低到0.6次/d。本文提出的方法表明,将人行为的建模分析与实际建筑运行控制相结合,可进一步改善控制效果。
Intelligent operation control of commercial buildings has been extensively studied in recent years,among which lighting system is one of the most significant systems.The control for lighting system in most commercial buildings is commonly based on real-time occupancy data collected by the passive infrared( PIR) sensor.The existing lighting control system adopts a time delay control approach.However,the detection inaccuracy of the PIR sensor usually results in a"false off"even though the room is still occupied.This research proposed a novel timinganalysis-based model predictive control( MPC) method using historical indoor occupancy state to overcome this challenge by predicting actual occupancy state using historical occupancy probability and time sequence features.The results of comparison between the proposed model control and traditional lighting system control indicated that,during test operation,the new method achieved a higher occupancy prediction accuracy of 97.4% and the "false off"frequency was reduced from 79.5 with traditional time delay method to 0.6 times per day.This method showed that combination of occupant behavior routine analysis and on-site building operation control can further enhance the control performance.
作者
晋远
燕达
安晶晶
李进
孙红三
吴如宏
杨一帆
JIN Yuan;YAN Da;AN Jingjing;LI Jin;SUN Hongsan;WU Ruhong;YANG Yifan(Tsinghua University,Beijing 100084,China;Zhuhai Singyes Green Building Technology Co.,Ltd,Zhuhai 519000,Guangdong,China)
出处
《建筑科学》
CSCD
北大核心
2020年第4期191-198,共8页
Building Science
基金
国家重点研发计划“净零能耗建筑关键技术研究与示范”任务四(2016YFE0102300-04)。
关键词
照明系统
人行为
模型预测控制
lighting system
occupant behavior
model predictive control
