MULTIOBJECTIVE OPTIMIZATION OF SEISMIC STRUCTURES

This paper presents a constrained multiobjective optimization method in the form of a robust, practical, problem-independent formulation based on genetic algorithm combined with game theory, and investigates the effect of multiobjective optimiztion on structural design. The study includes objectives, constraints, and design variables as well as their effect on structural design and behavior. The algorithm comprises – multiobjective fitness function, niche method, and Pareto set filter. A 10-story setback building is used to illustrate the evaluation of three choices: steel frame, reinforced concrete frame, composite steel and reinforced concrete frame. Evaluation is based on economics as well as performance. Comparison of optimum design results is obtained with consideration of weight, structural cost, and seismic energy for all three frames subjected to the same seismic input. Constraints are based on Uniform Building Code specifications including stress, displacement, drift, and ratio of story stiffness. The new approach provides a powerful tool to locate a global solution. Overview of Multioptimization Problemsolving In this research, game theory, fuzzy set theory, and Pareto genetic algorithm are applied to multiobjective optimization programming, constraint space transformation, and generation of nondominated sets. Game theory can find a compromise solution which satisfies all competing objectives in a cooperative game for a multiobjective optimization problem. Using fuzzy set theory can transform a constrained optimization problem into a nonconstrained one which can be optimized by a genetic algorithm. Pareto GA can generate a nondominated set even for