CAREER: Enhancing Conceptual Design Using Multi-Objective, Dynamically Encoded Genetic Algorithms to Optimize Structural Toplogy, Geometry, and Size

Texas A&M Engineering Experiment Station, College Station TX

Investigators

Abstract

This CAREER award supports research efforts focused on assisting designers in exploring and exploiting design alternatives during conceptual design. A conceptual design system will be developed that can be teamed with designers to create, evaluate, and optimize diverse design alternatives. A design grammar for representing complex, large-scale roof system design alternatives will be developed in collaboration with professional engineers and implemented using a specialized genetic algorithm (GA) that dynamically encodes the design variables. To assist in exploring design alternatives, a GA-based multi-objective (MO) method will be developed that obtains the s-Pareto frontier within a single trial for designs having different structural configurations. To select the best design from the s-Pareto set, a Kohonen neural network will be used to capture the user's design preferences. This research will significantly enhance a designer's ability to design efficient and novel structures and will influence the way designers conceptualize the design process. The CAREER educational activities seek to provide students with the skills and experience desired by their future employers. A new senior capstone design course will allow architectural and civil engineering students to collaborate on realistic problems defined by professional engineers. A new graduate course in heuristic methods will meet the demand for training interdisciplinary engineers to solve complex optimization problems. In addition, experiential learning modules will be added to undergraduate courses. The educational activities benefit students by increasing their knowledge of design and optimization, strengthening their understanding of structural behavior, and better preparing them to become future design innovators.