This paper discusses a new research effort that focuses on genetic algorithms as a tool for optimizing large-span roof trusses. The objectives to be minimized are truss weight, deflection, and fabrication cost. Decision variables include overall truss topology, joint geometry, and member sizing. A multi-objective genetic algorithm using an implicit redundant representation will be employed. Additionally, this research will incorporate user-feedback, in the form of penalty constraints, to guide the selection of aesthetically appealing design alternatives. In recent years, traditional optimization methods have been supplemented by heuristic algorithms that strive to blend mathematical models with expert knowledge. One such emerging method is the genetic algorithm, an optimization technique with roots in the theory of evolution and survival of the fittest. This paper details a new research effort to use genetic algorithms (GAs) in the design of large-span roof trusses. The long-term research goal is to create a program for generating more efficient designs while reducing project costs. This paper presents background material on GAs, a description of the problem, and the proposed research methodology.
展开▼