论文部分内容阅读
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 order to solve the problem that traditional genetic algorithm can not get satisfactory control parameters when the LQR controller of inverted pendulum is designed, the algorithm will be multi-population genetic algorithm with Sheffield genetic toolbox. At the beginning of the algorithm, multiple populations are introduced to search simultaneously. Different crossover probability and mutation probability are used in different populations. Migrants are used to establish contact between populations. The best individual generated by each population is retained by the elite population. In the Simulink environment, the model is established and the performance index is calculated. The output value under the Simulink environment is used as the objective function value of the multi-population genetic algorithm. The simulation results show that the proposed algorithm has good stability and low genetic algebra in the presence of modeling error, which can effectively avoid the premature problem and is more suitable for the optimization of complex problems.