中文摘要：

针对基本PSO算法在全局优化中易陷入局部极值和收敛精度低的不足,分析了基本PSO算法早熟收敛的原因,提出具有自适应邻域探测机制的改进型粒子群优化（ANE-PSO）算法.该算法在进化过程中以概率总体递减的方式,选择部分粒子对最佳位置按半径总体递减的规则进行邻域探测,并引入速度变异算子,提高种群的多样性,增强了算法的全局搜索能力.并证明它依概率1收敛到全局最优解.通过与其它三个改进算法比较,结果表明ANE-PSO具有较好的全局搜索能力,收敛速度较快,稳定性较好,且没有增加时间复杂度,较有效的避免了早熟收敛问题.

英文摘要：

In global optimization,particle swarm optimization（PSO） is often trapped in local optima and low accuracy in convergence.Following an analysis of the cause of the premature convergence,a novel particle swarm optimization based on self-adaptive neighborhood explored is proposed,which is called ANE-PSO.During evolution,every particle can explore the best position＇s neighborhood in a descend probability,the neighborhood radius can be self-adaptive to reduce,and also the velocity of mutation operator is added in.This method can break away from local optimization and enhances the global search ability.The ANE-PSO is guaranteed to converge the global optimization solution with probability one.Compared with other three improved algorithms on accuracy and convergence speed,and also on time complexity,it shows that the ANE-PSO converges faster,results in better optima,is more robust and the time complexity is not added,and prevents more effectively the premature convergence problem.