Particle Swarm Optimization in Structural Design
E. Ruben, Kamran Behdi · 2007
Particle Swarm Optimization is a population-based algorithm, which mimics the social behaviour of animals in a flock. It makes use of individual and group memory to update each particle position allowing global as well as local search optimization. Analytically the PSO behaves similarly to a traditional line-search where the step length and search direction are stochastic. Furthermore, it was shown that the PSO search strategy can be represented as a discrete-dynamic system which converges to an equilibrium point. From a stability analysis of such system, a parameter selection heuristic was developed which provides an initial guideline to the selection of the different PSO setting parameters. Experimentally, it was found that using the derived heuristics with a slightly larger cognitive pressure value