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

A computer-implemented method for generating a building operational performance analysis output comprising:

• (a) synthetically algorithmically generating, in a computing system using generative design techniques, a synthetic data set comprising a set of three-dimensional (3D) building conceptual mass geometries, and wherein the synthetic data set does not represent actual building forms, wherein the generating comprises:
• (i) identifying, in the computing device, two or more geometry types;
• (ii) dividing, in the computing device, the two more geometry types into categories at a footprint level and a massing level;
• (iii) algorithmically generating, in the computing device, the 3D building conceptual mass geometries using a separate workflow for each category, wherein the algorithmically generating utilizes the generative design techniques;
• (b) generating, in the computing device, an analytical model associated with each of the 3D building conceptual mass geometries;
• (c) generating, in the computing device, simulation results for each of the analytical models by conducting a simulation based on each of the analytical models;
• (d) training, by the computing device, a surrogate machine learning (ML) model based on a set of features, wherein the set of features is extracted from the simulation results using ML, wherein the ML iteratively determines the set of features, and wherein the surrogate ML model replaces an execution of simulations on proposed designs to reduce computational cost and simulation runtime thereby improving a functioning of the computing device in facilitating near real-time energy evaluations, and wherein the surrogate model uses the synthetic data set to obtain designs for specific building mass geometries;
• (e) receiving, in the computing device, geometry input;
• (f) processing, in the computing device, the geometry input through the surrogate model to generate the building operational performance analysis output;
• and
• (g) outputting, via the computing device, the building operational performance analysis output derived from the surrogate model, to provide a real-time energy prediction service based on geometric modeling operations, wherein the real-time energy prediction service integrates energy evaluation in a geometric modeling environment, wherein the building operational performance analysis output comprises an approximate Energy Use Intensity of the geometry input.