When thinking on retrofitting a building, energy models are used to estimate energy consumption for the different options. However, the accuracy of the model may vary depending on the accuracy of the input parameters. To minimize the uncertainty of the input parameters a calibration process is needed. As a mean to reduce the time for calibration, a methodology using a regression analysis with parameters or independent variables that are a function of the physical driving forces for energy performance is proposed. The main novelty of the approach is that it can be considered a quasi-physical statistical approach that has advantage over pure traditional statistical approaches. This is achieved by incorporating the physical driven forces of the parameters contributing to the energy consumption. Other advantage is that it uses the well know and easy to implement multiple regression analysis and system of equations. The approach also can be implemented in stages based on a prioritized list of parameters; that is, a few parameters with more impact are first used for a coarse tuning, and then additional parameters are used for fine tuning. The approach is applied by running a series of simulations with the design input parameters to obtain data for a multiple regression analysis. Then, the regression coefficients are used in a system of equations which solution gives the magnitude of the parameters. The results of the methodology were compared with the traditional Bayesian approach as a mean to validate the approach. Results show that the proposed approach is as accurate as the Bayesian approach.

在考虑改造建筑物时，能源模型用于估计不同选项的能源消耗。但是，模型的准确度可能会因输入参数的准确度而异。为了最小化输入参数的不确定性，需要一个校准过程。作为减少校准时间的一种方法，提出了一种使用回归分析的方法，其中参数或自变量是能量性能的物理驱动力的函数。该方法的主要新颖之处在于它可以被视为一种准物理统计方法，它比纯传统统计方法具有优势。这是通过结合对能耗有贡献的参数的物理驱动力来实现的。另一个优点是它使用了众所周知且易于实现的多元回归分析和方程组。该方法也可以根据优先级参数列表分阶段实施；即先用几个影响较大的参数进行粗调，然后再用附加参数进行微调。通过使用设计输入参数运行一系列模拟来应用该方法，以获得用于多元回归分析的数据。然后，在方程组中使用回归系数，该方程组的解给出了参数的大小。将该方法的结果与传统的贝叶斯方法进行比较，以此作为验证该方法的手段。结果表明，所提出的方法与贝叶斯方法一样准确。