基于实际运行数据校准居住建筑热负荷模型中的人行为时间表*
Calibration of occupant behavior schedule in residentialbuilding thermal load model with actual operation data
摘要：
建筑能耗模拟在建筑设计、运行和改造过程中发挥着重要作用，为了使其贴合实际，需要对其校准。作为影响建筑能耗的重要因素，建筑人行为模式的合理设置是能耗模拟结果准确的前提。提出了一种居住建筑热负荷模型的人行为时间表校准方法：基于建筑能耗模拟软件TRNSYS和最优化软件GenOpt联合仿真的半自动校准平台，使用k-means聚类方法对室外温度进行聚类，根据聚类结果将供暖季划分为不同的时间段，对不同时间段内的人行为时间表分别进行校准。将该方法应用于一栋居住建筑，其校准后模型模拟的误差评价指标MBE和CV（RMSE）从校准前的-81.71%和83.89%分别降低到-2.91%和15.90%，满足相关标准对模型校准精度的要求。
Abstract：
Building energy simulation plays an important role in the design, operation and retrofit of buildings. In order to make it realistic, it is necessary to be calibrated. As an important factor affecting building energy consumption, a reasonable occupant behavior schedule is a prerequisite of accurate energy simulation results. Proposes an occupant behavior schedule calibration method for residential building thermal load simulation. That is, based on the semi-automatic calibration method coupling TRNSYS with GenOpt, uses the k-means clustering method to cluster outdoor temperature; and according to the clustering result, divides the heating season into different time periods and calibrates the schedules in different time periods respectively. Applying this method to a residential building, the MBE and CV(RMSE) of the calibrated model are reduced from -81.71% and 83.89% to -2.91% and 15.90%, respectively, meeting the requirements of related standards.
Keywords:buildingthermalload,modelcalibration,occupantbehaviorschedule,energyconsumption,residentialbuilding