Evaluation of Energy Saving of Residential Buildings in North China Using Back-Propagation Neural Network and Virtual Reality Modeling

Abstract

Virtual reality (VR) modeling has been increasingly applied in the field of construction; however, research and progress in the design and evaluation of energy-saving building envelopes of residential structures in northern cold regions remain limited. The objective of this study is to identify less laborious methods to analyze the design of energy-efficient buildings using a back-propagation neural network (BPNN) combined with VR modeling to evaluate the building envelope structure. We first use the BPNN to construct an algorithm to calculate the heat transfer coefficient of the building envelope. Houses in Harbin are used as research objects, and a VR platform is used to construct architectural models of different envelopes. Results show that the error of the BPNN algorithm applied to the heat transfer coefficient identification of the building envelope is less than 5%, and that the VR software can realize the three-dimensional modeling of different energy-saving envelope structures, as well as facilitate the evaluation of energy-saving performance. The results can provide a theoretical basis for designers and decision-makers in the application of the BPNN and VR technology to the design and evaluation of building energy conservation.