Building Energy Model (BEM) calibration is the process of reducing the gap between the simulation outputs and the actual measured data at the same conditions. The literature shows that BEM calibration approaches could lead to a significant error in the model inputs even the calibration has been conducted successfully based on the model outputs (i.e., error functions). This paper compares the performance (i.e., accuracy and robustness) of 60 optimization-based calibration approaches. The approaches have different error functions (individual or combination of NMBE, NME, CV(RMSE), R² ,Cχ²) to be minimized and different outputs (heating demand, cooling demand, and\or indoor temperature for weeks, months, or a year) to be calibrated. The BESTEST600, predefined by ANSI/ASHRAE 140-2001, is selected as a white-box BEM case study for conducting the comparison test. Having the case study inputs and outputs without uncertainty gives a trustworthy comparison between the tested approaches. EnergyPlus is used for conducting the simulation while the Multi-objective optimization algorithm (a variant of NSGA-II) from MATLAB is used to minimize the error function(s) associated to each calibration approach. Among the 60 calibration approaches, eight proved to be the most accurate in predicting all calibration variables with percentage errors lower than 10%. CV(RMSE) was found to be the most robust error function under different calibration datasets. The results also show that the current standard calibration requirements are not proper as stopping criteria for automatic optimization-based calibration.

建筑节能模型（BEM）校准是缩小模拟输出与实际条件下的实际数据之间的差距的过程。文献表明，即使已经基于模型输出成功进行了校准（即误差函数），BEM校准方法也可能导致模型输入中的重大错误。本文比较了60种基于优化的校准方法的性能（即准确性和鲁棒性）。该方法具有不同的误差函数（NMBE，NME，CV（RMSE），R的个人或组合²，Cχ ²）以最小化，并校准不同的输出（供热，制冷和/或室内温度持续数周，数月或一年）。选择了由ANSI / ASHRAE 140-2001预定义的BESTEST600作为白盒BEM案例研究来进行比较测试。在没有不确定性的情况下进行案例研究的输入和输出，可以很好地比较测试方法。使用EnergyPlus进行仿真，同时使用MATLAB的多目标优化算法（NSGA-II的一种变体）来最小化与每种校准方法相关的误差函数。在60种校准方法中，有8种被证明是最准确的预测所有校准变量的方法，其百分比误差低于10％。发现CV（RMSE）是不同校准数据集下最强大的误差函数。结果还表明，当前的标准校准要求不适合作为基于自动优化的校准的停止标准。