In order to overcome the shortcomings of the standard particle swarm optimization algorithm such as premature and bad convergence, an improved particle swarm optimization algorithm based on dimension neighbors diffusion is pro-posed. The improved algorithm initializes to use the k-means to cluster the dimension of the particles in the space. The standard of the clustering is the Euclidean distance between each dimension. In the algorithm, each initial class from the cluster is considered as one family. The interior and exterior of the home are updated iteratively, and the idea of intelligent single-particle optimization algorithm is used to divide the update speed of the particles into the corresponding home speed subvectors. The diffusion and propagation of inter-particle exchange records draws upon the online social network propaga-tion model. In the process of adopting information, not only the value of information but also the surrounding particle status will be considered. Tabu search algorithm is used to establish the tabu list, set tabu search length and aspiration criterion to avoid the repeated search and improve the global search performance of the algorithm in order to increase the accuracy of the solution. The experimental results show that the improved algorithm can effectively solve the problem of slow convergence speed and low precision.%针对标准粒子群算法存在容易早熟及陷入局部最优等缺陷,提出了一种基于维度近邻关系扩散的改进粒子群优化算法.改进后的算法初始化时借鉴k-means对空间中粒子的维度进行聚类,聚类的标准为每个维度之间的欧氏距离,算法中将聚类得到的每个起始类视为一个家庭.家庭内部和外部分别进行迭代更新,结合智能单粒子优化算法的思想将粒子的更新速度划分为对应的家庭速度子矢量.粒子间交换记录的扩散和传播借鉴在线社会网络传播模型,在采纳信息的过程中不仅会考虑信息的价值,也会考虑其周围粒子状况.结合禁忌搜索优化算法,通过将该算法中的建立禁忌表、设定禁忌搜索长度和特赦准则等策略来避免重复搜索和改进算法的全局搜索性能,提高解的精确性.实验结果表明,改进后的算法有效解决了算法收敛速度慢、求解精度低等问题.
展开▼
