THE EFFECTS OF INITIAL SAMPLING AND PENALTY FUNCTIONS IN OPTIMAL DESIGN OF TRUSSES USING METAHEURISTIC ALGORITHMS

Although Genetic algorithm (GA), Ant colony (AC) and Particle swarm optimization algorithm (PSO) have already been extended to various types of engineering problems, the effects of initial sampling beside constraints in the efficiency of algorithms, is still an interesting field. In this paper we show that, initial sampling with a special series of constraints play an important role in the convergence and robustness of a metaheuristic algorithm. Random initial sampling, Latin Hypercube Design, Sobol sequence, Hammersley and Halton sequences are employed for approximating initial design. Comparative studies demonstrate that well distributed initial sampling speeds up the convergence to near optimal design and reduce the required computational cost of purely random sampling methodologies. In addition different penalty functions that define the Augmented Lagrangian methods considered in this paper to improve the algorithms. Some examples presented to show these applications.