Effective hybrid optimization strategy for complex functions with high dimension
Ling Wang · Journal of Tsinghua University(Science and Technology) · 2001
Classic optimization methods are easy to be trapped in local minima for complex functions with high dimension. This paper proposes an effective hybrid optimization strategy combining simulated annealing with simplex method, which simultaneously applies the probabilistic jumping search and the convex polyhedron based geometry search. The hybrid strategy is general, easily implemented and is of globally parallel optimization property. Numerical simulation results on benchmark complex functions with high dimension show that the hybrid strategy is effective, efficient, fairly robust to initial conditions and very adaptive to the dimensions. Especially the hybrid strategy is of strong ability to avoid being trapping in local minima, and its performances are fairly superior to those of single methods