Neural dynamics models for integrated optimum design of large steel structures.
详细信息
文摘
Neural computing models are developed for integrated optimum design of large steel structures with thousands of members subjected to the highly nonlinear constraints of actual design codes such as the American Institute of Steel Construction (AISC) Allowable Stress Design (ASD, 1989) or Load and Resistance Factor Design (LRFD) specifications (AISC, 1986 and 1994).;A neural dynamics model has been developed for optimal design of structures. The Lyapunov function is used to develop the neural dynamics structural optimixation model and prove its stability. An exterior penalty function method is adopted to formulate an objective function for the general constrained structural optimization problem in the form of the Lyapunov function. A learning rule is developed by integrating the Kuhn-Tucker necessary condition for a local minimum with the formulated Lyapunov function.;The integrated design consists of preliminary design, structural analysis, and the selection of the final members of the highrise or superhighrise structures made of commercially available W shapes. Finite element methods with various equation solving techniques and the adjoint variable method are used for structural analysis and sensitivity analysis.;A hybrid counter propagation-neural dynamics model and a new neural network topology has been developed for discrete optimization of large steel structures consisting of commercially available shapes and subjected to the AISC ASD specifications. The counter propagation network is trained to learn the relationship between the cross-sectional area and the radius of gyration of the available sections. The robustness of the hybrid computational model is demonstrated by application to three examples representing the exterior envelope of highrise and superhighrise steel building structures, including a 147-story structure with 8904 members. A robust data parallel neural dynamics model has been developed for discrete optimization of large steel structures based on the AISC ASD or LRFD specifications. The computational model has been implemented on a CM-5 supercomputer and applied to integrated minimum weight design of three steel highrise building structures ranging in height from 128.0 m (420 ft) to 527 m (1728 ft).