长沙市办公室空调使用行为分析与预测

Analysis and Prediction of Office Air-conditioning Use Behavior in Changsha

本文通过对长沙市某办公建筑内2间典型办公室的空调系统及室内外环境进行为期1 a的连续监测,研究办公人员空调使用行为特点并建立空调使用行为预测模型,为空调系统的节能与智能管控提供帮助。分析空调使用行为数据表明:大小办公室空调开启情况中分别有60.32%、90.58%发生在工作时间,且空调连续运行5~10 h的情况分别占比37%、54%;在供冷工况下,办公室空调设定温度集中在24~26℃;办公室空调风速档位大多设置为中档;大办公室夏季空调使用率和空调设定温度均普遍高于小办公室,冬季反之。同时,基于随机森林和XGBoost机器学习算法建立空调使用行为预测模型,结果表明:基于XGBoost算法的空调启闭行为预测模型准确率高达99%;基于随机森林算法的空调温度调节行为预测模型的Kappa系数最高为0.87。

Through the one-year continuous monitoring of the air conditioning system and indoor and outdoor environment of two typical offices in an office building in Changsha,this paper studies the characteristics of air-conditioning use behavior of office staff and establishes a prediction model of air-conditioning use behavior,which facilitates energy saving and intelligent control of the air-conditioning system.The analysis of air-conditioning use behavior data shows that 60.32% and 90.58% of the operating conditions of air-conditioning in large and small offices occur during working hours,and the continuous operation of air-conditioning for 5~10 hours accounts for 37% and 54%,respectively;In the cooling condition,the set temperature of office air conditioning is mostly 24~26 ℃;the fan gear of office air-conditioning is mostly set to mid-range.The utilization rate and set temperature of air conditioning in large offices are generally higher than those in small offices in summer,and vice versa in winter.At the same time,based on Random Forest and XGBoost machine learning algorithm,a prediction model of air-conditioning use behavior was established.The results show that the accuracy of the prediction model of air-conditioning opening and closing behavior based on XGBoost algorithm reached 99%.The highest Kappa coefficient of the prediction model of air conditioning temperature regulation behavior based on Random Forest algorithm is 0.87.