Building performance analysis (bpa) machine: machine learning to accelerate building energy analysis

Abstract

A method and system generate a building operational performance analysis output. A synthetic data set is generated and includes a set of 3D building conceptual mass geometries. The generating includes identifying geometry types, dividing the geometry types into categories, and algorithmically generating the mass geometries using a separate workflow for each category using generative design. Analytical models associated with each of the mass geometries are generated. Simulation results are generated for each of the analytical models. A surrogate model is trained based on a set of features extracted from the simulation results using machine learning (ML). The ML iteratively determines the set of features based on a measured accuracy of the surrogate model. Geometry input is received and processed through the surrogate model to generate the building operational performance analysis output which is then used to inform a designer of an approximate Energy Use Intensity of the geometry input.