As lots of the total energy was used by buildings, the number of residential buildings has dramatically increased in South Korea. Thus, it is imperative to pay more attention to energy consumption by residential buildings. In addition, it is important to predict energy consumption in residential buildings accurately. While several studies have currently focused on the data-driven method to predict energy consumption, it requires much information for multivariate data. The present study developed a predictive model for energy consumption for residential buildings by using the statistical method. Using the response surface methodology, the relationships between design factors, and heating and cooling energy use in residential buildings were outlined. The response values were calculated by using the simplified geometries in the energy simulation tool. The relationship has confirmed the dependencies of the energy consumption on various design variables of envelope systems in residential buildings in that the predictive model for the heating and cooling energy consumption was developed. The developed model was compared with the data obtained in the apartment buildings in two cities in South Korea. As a result, a coefficient of variation of the root mean squared error (Cv (RMSE)) was ranged from - 0.3% to 15% and all the comparisons were within the acceptable range. Moreover, heating and cooling energy consumption was predicted by varying the values of design variables such as thermal transmittance, solar heat gain coefficient (SHGC), and air infiltration rates. Among the variables, the largest heating energy was required as with the increase in the air infiltration rates, while the largest cooling energy was consumed as the SHGC was increased for both apartment buildings. Moreover, the increase in thermal transmittance values resulted in about 27% - 29% of the increase in the heating energy consumption. For cooling, 8% to 26% of the energy consumption was decreased when the thermal transmittance was increased. As can be shown, the developed model can offer a rapid energy prediction for apartment buildings with simple information on design variables. Furthermore, it can easily figure out the most important design factor to make a more energy-efficient residential building design.

由于建筑物消耗了大量的总能源，因此韩国的住宅建筑物数量急剧增加。因此，必须更加注意住宅建筑的能耗。此外，准确预测住宅建筑物的能耗也很重要。尽管目前有几项研究集中于数据驱动的方法来预测能耗，但对于多元数据它需要大量信息。本研究通过使用统计方法建立了住宅建筑能耗的预测模型。使用响应面方法，概述了设计因素与住宅建筑中供热和制冷能耗之间的关系。通过使用能量模拟工具中的简化几何来计算响应值。通过建立供暖和制冷能耗预测模型，该关系确定了能耗对住宅建筑围护系统各种设计变量的依赖性。将开发的模型与韩国两个城市的公寓楼中获得的数据进行了比较。结果，均方根误差的变异系数（Cv（RMSE））为-0.3％至15％，所有比较均在可接受的范围内。此外，通过改变设计变量的值（如热透射率，太阳热增益系数（SHGC）和空气渗透率）来预测加热和冷却的能耗。在这些变量中，随着空气渗透率的增加，需要最大的热能，随着两座公寓的SHGC的增加，消耗了最大的冷却能。而且，热透射率值的增加导致加热能量消耗的增加的大约27％至29％。对于冷却，当增加热透射率时，能量消耗减少了8％至26％。如图所示，开发的模型可以提供有关设计变量的简单信息的公寓建筑物的快速能源预测。此外，它可以轻松找出最重要的设计因素，以进行更节能的住宅建筑设计。当热透射率增加时，能量消耗的8％至26％减少。如图所示，开发的模型可以提供有关设计变量的简单信息的公寓建筑物的快速能源预测。此外，它可以轻松找出最重要的设计因素，以进行更节能的住宅建筑设计。当热透射率增加时，能量消耗的8％至26％减少。如图所示，开发的模型可以提供有关设计变量的简单信息的公寓建筑物的快速能源预测。此外，它可以轻松找出最重要的设计因素，以进行更节能的住宅建筑设计。