摘要
宇宙大爆炸算法(Big Bang-Big Crunch,BB-BC)思想来源于宇宙大爆炸和大收缩理论.针对其在高维函数的寻优过程中,随迭代次数增加,爆炸生成的碎片解收缩速度慢,多样性快速减弱,质量变差,容易陷入局部最优解的缺点,提出一种混合型BB-BC算法(HBB-BC).首先,将质心代入当代解中作为奇点解进行改进,提高算法收缩速度;其次,结合粒子群优化的路径优化,提高碎片解的质量;最后,引入宇宙大撕裂理论增加大爆炸阶段碎片解的多样性和跳出局部最优解的能力.通过9个新型测试函数进行测试,测试结果显示,HBB-BC算法在高维函数的寻优性能上更优于BB-BC算法和另一种改进的均匀大爆炸混沌大收缩(UBB-CBC)算法.
The Big Bang-Big Crunch （ BB-BC ） algorithm is based on the big bang and big contraction theory of the universe. With the increase of number of iterations in optimizing of high dimensional func-tions, the candidates shrink slowly, worsen in quality and weaken rapidly in diversity, as well as sink in-to a local optimal solution. In light of these features, an improved hybrid BB-BC algorithm （ HBB-BC） is proposed. This algorithm puts the center of mass into contemporary candidates computing as a singular point solution to increase the speed of contraction and improves the candidates’ quality and enhances its diversity by mean of Particle Swarm optimization （PSO）. At last, Big Rip theory is introduced to in-crease the diversity of the big bang phase solutions and the ability to jump out of local optimal solution. The experimental results tested by 9 new benchmark test functions indicate that the improved algorithm performs better than the BB-BC and Uniform Big Bang-Chaotic Big Crunch （ UBB-CBC） on optimization of high dimensional functions.
出处
《广东工业大学学报》
CAS
2016年第4期12-17,共6页
Journal of Guangdong University of Technology
基金
国家自然科学基金资助项目(61502108)
广东省重大科技专项资助项目(2014B010111007)
广东省自然科学基金资助项目(2014A030313512)
作者简介
吴伟民（1956-）,男,教授,硕士生导师,CCF会员（E200030580M）,主要研究方向为可视计算、系统工具与平台.
