DEEP-LEARNING BASED FUNCTIONAL CORRELATION OF VOLUMETRIC DESIGNS

摘要

A design application receives an exemplary design from an end-user having one or more functional attributes relevant to solving a design problem. The design application then generates a set of labels that describes the functional attributes of the exemplary design. Based on the set of labels, the design application explores a functional space to retrieve one or more system classes having functionally descriptive labels that are similar to the set of labels generated for the exemplary design. The one or more system classes include different approaches to solving the design problem, and represent systems having at least some functional attributes in common with the exemplary design.

主权项

1. A computer-implemented method for generating one or more design approaches that include one or more desired functional attributes, the method comprising:
generating, based on an exemplary design, a first set of labels that describe a first plurality of functional attributes of the exemplary design; determining, based on the first set of labels, a second set of labels that describe a second plurality of functional attributes of a first class of systems; and extracting, based on the second set of labels, the first class of systems from a functional space that includes a plurality of system classes, wherein each system in the first class of systems has at least one functional attribute directed towards solving the design problem.