Solving truss topological optimization with discrete design variables via swarm intelligence

Abstract

Structural topological optimization is the most general form of structural optimization and one of the most challenging research problems in this field is to find optimized layouts with minimization of compliance (maximization of stiffness) for a given total mass of the structure discretized by truss members in terms of discrete design variables, which has not been well solved by evolutionary algorithms. Particle Swarm Optimization (PSO) is a new paradigm of Swarm Intelligence which is inspired by concepts from ‘Social Psychology’ and ‘Artificial Life’. PSO is particularly a preferable candidate to solve highly nonlinear, non-convex and even discontinuous problems and has been applied to many different kinds of optimization problems. The motivation of this paper is to use the Modified Lbest based PSO (MLPSO) and geometrical consistency check tightly connecting to the ground structure approach to break through in this kind of optimization problems. In order to handle discrete design variables, rounding-off strategy is used in this paper. Furthermore, external quadratic penalty function is chose to deal with constraints. Through a popular benchmark test, two kinds of MLPSO exhibited competitive performance due to improved global searching ability.