Paper Conference

2020 Building Performance Analysis Conference and SimBuild co-organized by ASHRAE and IBPSA-USA


ARINet: Using 3D Convolutional Neural Networks to Estimate Annual Radiation Intensities on Building Facades

Jung Min Han, Chih-Kang Chang, Ali Malkawi
Harvard Graduate School of Design, Cambridge, MA
Harvard Center for Green Buildings and Cities, Cambridge, MA
Harvard John A. Paulson School of Engineering and Applied Sciences, Cambridge, MA

Abstract: Artificial intelligence and data-driven modeling are becoming more prominent in the building, and construction sectors. Physics-based models usually require significant computational power and a considerable amount of time to simulate output. Therefore, data-driven models for predicting the physical properties of buildings are becoming increasingly popular. The objective of this research is to introduce Artificial Neural Networks (ANNs) methods as a means of representing the physical properties of buildings. Achieving this goal will illustrate the future capacity of integrated neural networks in building performance simulations. The Annual Radiation Intensity Neural Network (ARINet) demonstrates the feasibility of using a 3D convolutional neural network to predict the surface radiation received by building façades. The structure of ARINet is composed of 3D convolution, fully connected, and 3D deconvolution layers. In this research, it was trained on 1,692 datasets and validated by 424 datasets generated by a physical simulator. ARINet showed errors in 0.2% of the validation sets.
Pages: 252 - 259