多种群遗传算法的倒立摆LQR控制器设计  

LQR Controller Design for Inverted Pendulum Based on Multi-population Genetic Algorithm

作　　者：谌海云[1] 杜振华[1] 邹宁波[1] 石明江[1] 

机构地区：[1]西南石油大学电气信息学院,成都610500

出　　处：《控制工程》2014年第3期391-394,共4页Control Engineering of China

基　　金：四川省教育厅重点科研项目(12ZA193)

摘　　要：针对传统遗传算法在设计倒立摆LQR控制器时,算法会因个别超常个体或群体较快趋于单一化而导致不能得到满意控制参数的问题,利用Sheffield遗传工具箱,设计了多种群遗传算法。算法的开始引入了多个种群同时进行优化搜索,不同的种群采用不同的交叉概率和变异概率,种群之间采用移民算子建立联系,各种群产生的最优个体通过精华种群实现留优。在Simulink环境下建立模型并计算性能指标,Simulink环境下的输出值作为多种群遗传算法的目标函数值。仿真结果表明,在存在建模误差的情况下,本算法稳定性好、遗传代数小,有效地避免了早熟,更为适合复杂问题的优化。In view of the premature convergence of traditional genetic algorithm that lead to the problems can not get satisfied parameters in designing LQR controller of the inverted pendulum, this paper introduce a multi-popula-tion genetic algorithm based on Sheffield ge- netic toolbox to overcome the fault that due to several supernormal in-dividuals or population tend to simplification too fast. The algo- rithm optimize the problem parallelly, the different populations use different crossover probability and mutation probability. The popula- tions establish inter - contact by using immigration operator, preserve the best individual of the various group in quintessence popula- tion. Simulink is used to model and then compute the performance indicators, the output value of simulink is used as the objective func- tion value of the multi - population genetic algorithm. Simulation results show that, in the presence of modeling errors, the algorithm is stability, generations is smaller and avoid the premature convergence effectively. This algorithm is more suitable for complicated problem optimization.

关 键 词：LQR控制器 多种群遗传算法 倒立摆 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]