摘要
标准群搜索优化算法易陷入局部最优。为此,引入模拟退火策略和差分进化算子,使算法跳出局部极值点,变异和迭代同时进行,并保持前期搜索速度快的特性。测试结果证明,改进算法的全局收敛能力明显提高,个体具有良好的人工智能性,能够真实模拟群体行为。
To reduce the possibility of falling into local optimum, metroplis rule and differential evolution operator is introdued in Group Search Optimization algorithm, which makes the variation of the algorithm get rid of the shackles of the local extreme advantage, maintain the pre-fast search feature, and improve global search capabilities. During varation and iterative, merit-based evolution improves optimization performance. Test results point that the ability to improve the global convergence of the algorithm is significantly improved, meanwhile, simulation results show that the individual has a good artificial intelligence, which can simulate group behavior toralistically.
出处
《计算机工程》
CAS
CSCD
2012年第17期178-181,共4页
Computer Engineering
基金
国家自然科学基金资助项目(6097004)
教育部博士点基金资助项目(20093704110002)
山东省自然科学基金资助项目(ZR2010QL01)
关键词
群搜索优化算法
群体动画
差分进化
局部最优
Group Search Optimization(GSO) algorithm
group animation
Differential Evolution(DE)
local optimum