摘要
提出并实现了一种新的全局最优化求解方法──Ｄａｒｗｉｎ＆Ｂｏｌｔｚｍａｎｎ混合策略，它是一个综合了Ｄａｒｗｉｎ进化策略和Ｂｏｌｔｚｍａｎｎ退火策略特点的通用随机迭代算法．理论分析证明，这一求解算法渐近收敛于全局最优解集且可具有多项式计算复杂性．实例计算结果表明，Ｄａｒｗｉｎ＆Ｂｏｌｔｚｍａｎｎ混合策略求解全局最优化问题好于现有方法，特别远优于模拟退火法．
Efficient global optimization techniques have extensive and important significances for theory and applications. This paper proposes and implements a new solving approach,the Darwin & Boltzmann mixed strategy, for global optimization problems. It is a general stochastic iterative algorithm which synthesizes the advantages of the Darwin strategy and the Boltzmann annealing strategy. The theoretical analyses prove that the algorithm converges asymptotically to global optimums and is a polynomial one. The results of experimental evaluations on the generally used problems show that the algorithm consistently performs as well as the existing techniques and even better, and specially, is superior enormously to the simulated annealing algorithm.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
1996年第11期20-27,共8页
Journal of Shanghai Jiaotong University
基金
中国博士后科学基金
中科院自动化所复杂系统工程学开放实验室资助
关键词
全局最优化
D&B混合策略
多项式算法
global optimization
Darwin & Boltzmann mixed strategy
global asymptotical convergence
polynomial algorithm
