拟人智能控制及鲁棒LQ控制在倒立摆基准问题中的应用

Application of human-imitating intelligence control and robust LQ control to the benchmark problem for the inverted pendulum

分别应用拟人智能控制策略解决倒立摆标称系统的控制和鲁棒LQ方法解决其鲁棒控制问题.拟人智能控制模仿人解决问题的归约思路,从物理角度出发分析被控系统并设计定性控制律.利用遗传算法良好的全局搜索收敛特点,对定性控制律中的参数进行优化搜索.当模型只存在结构化型不确定性且不确定性有界时,可通过求解一个Riccati方程来设计鲁棒LQ控制器.仿真结果表明给定的控制指标均得到满足,且控制律算法简单,实现比较方便.

Human-imitating intelligence control method and robust linear quadratic(LQ) method are applied to the nominal and uncertain inverted pendulum system respectively.The former method analyzes the (plant) to be controlled and derives the (qualitative) control law from the point of view of physics by adopting the general reduction method.The coefficients of the (qualitative) control law can be searched by genetic algorithm,which has globally convergent property.As for uncertain systems in state space representation,if the parameter uncertainty is structured and value bounded,the robust LQ controller is designed by resolving one Riccati equation.The simulation shows that these methods are effective and efficient in dealing with the inverted pendulum benchmark problem.