Enhanced Colliding Bodies Algorithm for Truss Optimization with Frequency Constraints

Abstract

To improve the performance of the structure and keep out the resonance phenomenon, some limitations should be imposed on the natural frequency ranges. This paper illustrates the computational performance of enhanced colliding bodies optimization (ECBO) in optimum weight design of truss structures with multiple natural frequency constraints. The standard CBO is a multiagent algorithm based on one-dimensional collisions between two bodies. Each agent is modeled as the body with a specified mass and velocity. The enhanced CBO utilizes memory to save some historically best solutions and also uses a mechanism for jumping out of local minima. The viability and efficiency of the proposed method are demonstrated by using four truss structures subject to constraints on natural frequencies. In frequency constraint problems, the search space is highly nonlinear and nonconvex with numerous local optima. Numerical results are compared with standard CBO results and results reported in the literature. These results indicate that the ECBO is a powerful method for optimum design.