Particle Swarm Optimization in Structural Design

Abstract

Optimization techniques play an important role as a useful decision making tool in the design of structures. By deriving the maximum benefits from the available resources, it enables the construction of lighter, more efficient structures while maintaining adequate levels of safety and reliability. A large number of optimization techniques have been suggested over the past decades to solve the inherently complex problem posed in structural design. Their scope varies widely depending on the type of stru ctural problem to be tackled. Gradient-based methods, for example, are highly effectively in finding local optima when the design space is convex and continuous and when the design problem involves large number of design variables and constraints. If the problem constraints and objective function are convex in nature, then it is possible to conclude that the local optimum will be a global optimum. In most structural problems, however, it is practically impossible to check the convexity of the design space, therefore assuring an obtained optimum is the best possible among multiple feasible solutions. Global non-gradient-based methods are able to traverse along highly non-linear, non-convex design spaces and find the best global solutions. In this category many unconstrained optimization algorithms have been developed by mimicking natural phenomena such as Simulated Annealing (Kirkpatrick et al., 1983), Genetic Algorithms (Goldberg, 1989), and Bacterial Foraging (Passino, 2002) among others. Recently, a new family of more efficient global optimization algorithms have been developed which are better posed to handle constraints. They are based on the simulation of social interactions among members of a specific species looking for food sources. From this family of optimizers, the two most promising algorithms, which are the subject of this book, are Ant Colony Optimization (Dorigo, 1986), and Particle Swarm Optimization or PSO. In this chapter, we present the analysis, implementation, and improvement strategies of a particle swarm optimization suitable for constraint optimization tasks. We illu strate the functionality and effectiveness of this algorithm, and explore the effect of the different PSO setting parameters in the scope of classical structural optimization problems.