摘要
为改善免疫网络算法在多峰函数优化方面存在局部收敛的不足,提出一种Lamarck免疫网络算法(LM-aiNet)。依据Lamarck进化理论思想,设计基于方向的局部搜索机制和自适应的网络抑制阈值,提高了算法对于不同类型多峰函数优化的适应能力。对算法的复杂度和收敛性进行分析,重点讨论了算法主要参数对求解性能的影响,确定合适的参数取值范围。实验结果表明,算法能够有效地解决经典算法的局部收敛问题,其求解能力优于被比较的其他算法。
In order to solve the problem of local convergence in immune network algorithm, this paper presented a Lamarck immune network algorithm（ LM-aiNet）. According to Lamarck evolution theory, this paper designed a directional local search strategy, and added an adaptive network suppress threshold to the algorithm. As a result, new algorithm is more adaptable for different multimodal functions. This paper performed the computing complexity and convergence analysis. It studied the influence of important parameters on algorithm＇ s performance and determined the suitable range of parameters. Results on benchmark functions show that LM-aiNet avoids the problem of local convergence effectively. And its searching performance is better than the other compared algorithms.
出处
《计算机应用研究》
CSCD
北大核心
2012年第3期902-906,共5页
Application Research of Computers
基金
航空科学基金和航空电子系统综合技术国防科技重点实验室联合资助项目(20095584006)
山东省自然科学基金资助项目(Y2008E11)
