基于分段遗传算法的倒立摆摆起控制研究

On Inverted Pendulum Swing-up Control Based on Genetic Algorithm

针对倒立摆摆起这类复杂非线性系统的优化控制问题,提出了一种基于遗传算法的近似全局最优化的控制方法。在导轨长度和控制力受限的情况下,从倒立摆的物理特性出发,以摆杆进入上半平面前的最后一次角速度为零的时刻为分界时刻,将摆起过程分成两个阶段,分别使用自适应交叉、变异算子的遗传算法搜索每个阶段的最优控制决策,实现了1.9 s内一级倒立摆的成功摆起。这种方法可以简化控制规则,优化摆起时间,并使摆杆更加平稳地进入稳摆控制。仿真和实时控制实验,说明该方法是可行的。

To the optimal control problem of the inverted pendulum swing up,an approximate global optimization control method is put forward based on the genetic algorithm.Under the limitation conditions of rail length and control force,the dividing moment is the final time when the rotational speed is zero before the swing stem entered in the upper half plane,and the pendulum swing-up course is divided into two stages according to the physical character of the pendulum.The genetic algorithms with self-adapting cross operators and variation operators are applied to research the optimal control decision in each stage respectively.The single inverted pendulum succeeds in swing-up within 1.9 s.This method could simplify the control rule,optimize the swing up time and put the swing stem in a more stable swing control state.The simulation and real-time experiments show that the method is feasible.