高效动态微型多目标遗传算法及其应用  

Efficient Dynamic Micro Multiobjective Optimization Method and Its Application

作　　者：刘桂萍[1] 于瑞贞 

机构地区：[1]湖南大学机械与运载工程学院,湖南大学汽车车身先进设计制造国家重点实验室,长沙410082

出　　处：《系统仿真学报》2014年第2期260-266,共7页Journal of System Simulation

基　　金：国家自然科学基金(11202073);湖南省自然科学基金(12JJ4008)

摘　　要：针对具有多个优化目标且目标和约束会随时间(环境)变化的动态优化问题,提出了一种高效的动态多目标遗传算法。该算法在微型遗传算法的基础上,针对动态优化问题的特点,加入一种环境检测机制,以实现对不同环境下的Pareto最优解集的快速求取。通过对四种不同类型的动态多目标优化测试问题的求解,并与经典算法DNSGAII进行对比,验证了该算法具有较高的求解效率和求解精度。最后,将该算法应用于一个动态的垃圾焚烧系统的PID控制参数的优化问题中,将阶跃响应下的最大超调量和上升时间作为优化目标,对PID比例系数和微分系数两个参数进行优化,结果表明,算法能够快速求出不同环境下的Pareto最优解集。An efficient dynamic multiobjective genetic algorithm based on the micro genetic algorithm was suggested to solve dynamic multiobjective optimization problems, which optimization objectives and constraints changed over time （Environment）. An environmental detection mechanism was employed to efficiently obtain the Pareto optimal sets of different environments of the dynamic optimization problem. Simulation results for several difficult test functions indicate that the present method has higher efficiency and better convergence near the globally Pareto-optimal set for all test functions, and a better spread of solutions for some test functions compared to non-dominated sorting genetic algorithm II. Eventually, this approach is applied to the PID control of a dynamic refuse incineration system for minimum the maximum overshoot and rise time of step response.

关 键 词：动态优化 多目标遗传算法 微型遗传算法 PID控制 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]