Coevolutionary approaches to structural optimization

A coevolution method for solving a large class of optimization problems is introduced. The new method solves static zero-sum games or saddle-point problems by an artificial evolution of two groups with opposite objectives. The 72-bar truss problem is used as an example model to verify the proposed coevolution method. Numerical results show that the coevolution method can achieve the global solution with an accuracy comparable to that of conventional nonlinear programming techniques. Further, the coevolution method is robust to initial conditions, producing consistent solutions.