人行为模型反映特性的比较研究.PDF

增刊 空调 建 筑 技 术 开 发 2016 11 Air conditioner Building Technology Development 年 月 人行为模型反映特性的比较研究 简毅文，高 萌，刘昭辉，常小艳 100124 （北京工业大学建筑工程学院，北京 ） ［摘 要］针对建筑人行为随机模型中的 Logistic 回归模型和清华大学提出的基于条件触发的动作模型，开展 两者在人行为特性定量描述上的分析对比研究。以北京某高校学生宿舍分体空调的开、关行为为分析案例，基于 房间温度、湿度的现场测试数据、室内空调运行的问卷调查信息及由此统计得出的不同驱动状况下空调开、关动 作发生概率的离散数据，分别应用 Logistic 回归模型与动作模型拟合确定并分析比较各个驱动力下宿舍空调开、关 行为的特征参数及空调开、关动作发生概率的变化趋势。研究结果反映出两个模型各自的应用优势，同时表明在 准确得到离散点概率统计值的情况下，Logistic 回归模型与动作模型都可以对居住建筑间歇运行的空调行为给予真 实的描述。 Logistic ［关键词］ 回归模型；动作模型；动作发生概率；特征参数 Comparative Study on Characteristics of Human Behavior Model Jian Yi-wen, Gao Meng, Liu Zhao-hui, Chang Xiao-yan According to the Logistic regression model and the action model based on conditional triggers proposed ［Abstract］ by Tsinghua University, a comparative study of quantitative analysis on the characteristics of human behavior has been car- ried out. Based on case studies which focuses on the AC usage at a university dormitory, the discrete data of ON/OFF sched- ule probability of AC drived by different triggers has been concluded with measurements of the indoor temperatures,hu- ， midity,and some spot interviews about AC usage patterns. Logistic regression model and action model are used to match fitting curves respectively which can determine and compare the characteristic parameters of ON/OFF action of split AC ， drived by different triggers and variation tendency of ON/OFF action probability of AC. The studies show that two models possess advantages respectively. On the condition that the measured data points are accur