模拟退火算法的一种改进及其应用研究

An improvement of simulated annealing algorithm and study of its application

针对连续函数全局优化问题提出了改进的模拟退火算法:采用新的解扰动策略,并将局部极小化过程引入模拟退火算法.数值试验证实了该算法的可行性及有效性.作为该方法的应用,计算了著名的L ennard-Jones簇问题,通过比较说明新方法可以提高精度及成功率;此外对脑啡肽的空间结构进行了预测,也得到了较好的结果.

The improved simulated annealing algorithm is introduced for global optimization of continuous function. The new method to generate the next point is adopted and the local minimum is added to the simulated annealing algorithm. Numerical results indicate that this algorithm is feasible and effective. As applications of this method, the famous problem of Lennard-Jones cluster is computed. The results show that the new method can improve the precision and the percentage of success. Moreover the 3-D structure of met-enkephalin is predicted and the better results are obtained.