Improved adaptive simulated annealing genetic algorithm (GA)

Simulated annealing GAs suffer from shortcomings such as insufficient search efficiency and premature convergence. We now propose an improved adaptive simulated annealing GA that possesses better search efficiency and the capability to converge to good global optimum even for high-dimensional complex functions. The description of traditional simulated annealing GAs and proposed adaptive simulated annealing GA and the six characteristics of traditional simulated annealing GAs are described. The proposed adaptive simulated annealing GA is described and cross probability and mutation probability of the proposed algorithm are selected adaptively for enhancing algorithm stability and convergence. The proof of our theorem for the convergence of the proposed adaptive simulated annealing GA is also presented, which is rather lengthy and takes up more space. Finally, for comparing our proposed algorithm with traditional simulated annealing GAs and the improved evolutionary programming algorithm, we give a numerical simulation example. These results demonstrate the effectiveness and efficiency of the proposed algorithm as applied to high-dimensional complex functions and its performances are better than those of traditional simulated annealing GAs and the improved evolutionary programming algorithm.