自适应混合演化算法

自适应混合演化算法 Self-Adaptive Hybrid Evolutionary Algorithms 高飞 GAO Fei 1 1 2 first-author 钟守楠 ZHONG Shou\|nan 1 1 2 纪昌明 JI Chang\|ming 2 1 2 武汉大学数学与统计学院 武汉大学数学与统计学院 School of Mathemetics and Statistics, Wuhan University School of Mathemetics and Statistics, Wuhan University 武汉大学水利水电学院,湖北,武汉,430072 武汉大学水利水电学院,湖北,武汉,430072 School of Water Resources & Hydropower, Wuhan University School of Water Resources & Hydropower, Wuhan University 为避免标准的演化算法演化速度慢、易收敛到局部极值的缺点 ,融合对梯度的随机模拟、免疫算子、模拟退火算法的思想 ,提出一种自适应混合演化算法 ,它在不同的演化阶段自适应的采用不同的演化算子 ,在演化初期具有较强的全局搜索性能 ,在演化中后期具有较强的精搜索性能 ,能迅速收敛于全局最优解 ;对标准测试函数的仿真结果表明 ,该算法具有精度高 ,收敛速度快 ,稳健性强的优点 A new Self\|adaptive hybrid evolutionary algorithm （SHEA） is presented in this paper based on other researchers’ work by adding some ideas of Monte Carlo Simulation of gradient, immune operator and simulated annealing to Evolutionary algorithms, which is to avoid Simple Evolutionary algorithms’ low evolution speed and premature. The new algorithm uses different evolution operators self\|adaptively during different evolution periods and it has high global searching performance in forepart of evolution and high local performance in late evolution, and it can converge to global optimum quickly. As the simulated experimental results show, Self\|adaptive hybrid evolutionary algorithm has the advantages of high precision and robustness and fast convergence. 自适应 演化算法 模拟退火 梯度的随机模拟 免疫算子 O224 国家自然科学基金资助 (70 1710 16;5 0 0 9962 4);亚太运筹学中心;中国科学院管理决策处信息系统开放研究室资助课题 2002-05-01 2021-04-01 5