摘要
给出了基于粒子群算法的组合预测方法,并引入免疫算法对其进行了改进:一方面利用免疫算法的免疫记忆和自我调节机制提高其全局搜索能力,避免算法陷入局部最优解;另一方面利用免疫算法的接种疫苗和免疫选择机制增强其性能,防止算法在优化过程中可能出现的退化现象。实例证明,基于免疫粒子群算法的组合预测方法可操作性强,通用性好,误差明显小于各个参与组合的预测模型,并优于基本的粒子群算法和加速遗传算法。
A combined forecasting method based on particle swarm optimization （PSO） is given in this paper, and it is also improved by immunity algorithms （IA）. On the one hand,memory and density mechanism of IA are used to enhance PSO＇s abillity of seeking the global excellent result ,avoiding PSO getting into local best place. On the other hand,vaccination and immune selection of IA are used to strengthen PSO＇s performance, preventing the undulate phenomenon during the evolutionary process. Two applications of PSO with IA （1A-PSO） are demonstrated in this paper. And they show that the combined forecasting method based on IA-PSO is very operable and general,and has evidently less predicted errors than those of each model,basic PSO and accelerating genetic algorithm.
出处
《系统工程理论方法应用》
北大核心
2006年第3期229-233,共5页
Systems Engineering Theory·Methodology·Applications
关键词
粒子群算法
免疫算法
组合预测
优化
particle swarm optimization
immunity algorithm
combined forecasting
optimization
