基于精英集聚效应的实数编码小世界优化算法

为了提高小世界算法的高维模型优化性能和降低算法的编码复杂性,提出了一种基于精英集聚效应的自适应实数编码小世界优化算法。该算法借鉴小世界现象进行网络空间搜索,包括随机长连接和局部短连接。为了提高优化性能,首先基于精英集聚效应在长连接中加入分级个体吸引策略;然后根据节点优化优劣在短连接中进行搜索次数及邻域大小的自适应调整。为了降低编码复杂性采用了实数编码。最后通过Markov链理论证明了算法的收敛性。数值测试结果表明,与禁忌遗传算法、基本小世界算法以及禁忌小世界算法相比,该算法在相对误差方面平均降低了30.3%,在收敛速度和稳定性方面分别平均提高了18.2%和13.8%,从而验证了算法的有效性。

Real Coding Small World Optimization Algorithm Based on Elite Gathering Effect

To improve the high dimensional model optimization performance of small world algorithm and reduce the complexity of algorithm coding, this paper proposes an Adaptive Real Coding Small World Algorithm based on elite gathering effect（ARCSWA）. Inspired by small world phenomena, the algorithm performs network search, including random long-range connection and local short-range connection. To improve the optimization performance, the hierarchical individual attraction strategy is firstly added to the long-range connection based on the elite gathering effect. Then, the number of searches and the size of the domain are adaptively adjusted in the short-range connection according to the node optimization. In addition, real coding is used to reduce coding complexity. Finally, the convergence of the ARCSWA is proved by Markov chain theory. The numerical test results show that, compared with the Tabu Genetic Algorithm（TGA）, the Simple Small World Algorithm（SSWA） and the Tabu Small World Algorithm（TSWA）, the average error of the proposed algorithm is reduced by 30.3% the convergence speed and stability of the proposed algorithm are increased by 18.2% and 13.8%, respectively, which verifies the validity of the ARCSWA.