摘要
微分进化算法作为一种新型、简单、高效的并行随机优化算法,近年来在许多领域得到了应用,多目标微分进化便是其中的一种。针对传统多目标微分进化算法中微分进化控制参数不能自适应调整、算法容易出现早熟和退化的现象,采用惯性权重参数自适应调整的控制策略以及改进的拥挤距离算法对多目标微分进化进行改进,并将改进后的算法用于控制系统PID参数优化仿真试验。结果表明,改进后的多目标微分进化算法具有较好的收敛性和分布性以及较高的搜索效率。
As a new type of simple and efficient parallel stochastic optimization method,the differential evolution algorithm has been applied in many fields in recent years and multi-objective differential evolution is just one of them.In traditional multi-objective differential evolution algorithm,the differential evolution control parameters cannot be adaptively adjusted,thus the algorithm is easy to be precocious and degraded.By adopting the control strategy of inertia weight parameter adaptive adjustment and the improved crowding distance algorithm,the multi-objective differential evolution is improved.The improved algorithm is used in simulation experiment for parameter optimization of PID control system.The result indicates that the improved algorithm features better convergence,distribution and high search efficiency.
出处
《自动化仪表》
CAS
北大核心
2012年第2期1-4,共4页
Process Automation Instrumentation
基金
国家863计划基金资助项目(编号:2007AA041106)
关键词
多目标
优化算法
微分进化
自适应
参数优化
PID
Multi-objective Optimization algorithm Differential evolution Self-adaptive Parameter optimization PID