A robust global optimization method based on simulated annealing

A robust global optimization algorithm based on simulated annealing is proposed in which global optima are searched for in regions that have high probability of containing optima and the domain of search is successively reduced until the stopping criterion is satisfied. By introducing the ideas of cumulative probability distribution function and stable energy, the selection of initial temperature and equilibrium criterion in the process of simulated annealing becomes easy and effective. Numerical studies using a set of standard test functions show that the proposed approach is effective and robust in solving global optimization problems.