双轮倒立摆机器人的模型预测控制策略  

Model Predictive Control Strategyof Two-Wheeled Inverted Pendulum Robot

作　　者：孔国利 张璐璐 KONG Guo-li;ZHANG Lu-lu(School of Information Engineering Zhengzhou Institute of Technology,He’nan Zhengzhou 450000,China)

机构地区：[1]郑州工程技术学院信息工程学院,河南郑州450000

出　　处：《机械设计与制造》2018年第9期265-268,共4页Machinery Design & Manufacture

基　　金：河南省科技攻关项目(182102210150)

摘　　要：由于欠驱动的双轮倒立摆机器人(Two Wheeled Inverted Pendulum Robot,TWIPR)是一个非线性和强耦合的不确定性系统,设计了一种利用过程显式模型优化系统性能的模型预测控制(Model Predictive Control,MPC)策略,采用了两个不同的模型预测控制器,并通过引入解耦单元实现了对双轮倒立摆机器人的独立控制。此外,采用前馈控制方法提高MPC控制器对可测性干扰的补偿,增强了MPC控制器的有效性和鲁棒性。最后,采用梯状干扰和两个不同的外力对设计的MPC控制器性能进行评估,并分别计算不同外部干扰下倾斜角和旋转角响应的均方根误差(Means Square Error,MSE),然后将其与线性二次调节(Linear Quadratic Regulator,LQR)控制器的控制性能进行对比。比较结果表明:MPC控制器的MES比LQR控制器均减小了50%以上,证明了MPC控制器对TWIPR的控制具有明显的优越性、可靠性和鲁棒性。In view of the Two Wheeled Inverted Pendulum Robot(TWIPR)is an uncertain system with serious nonlinearity and strong coupling,a model predictive control(MPC)strategy that uses an explicit model of a process to optimize the system performances is designed.By employing a decoupling unit,TWIPR can be controlled independently by two MPC controllers.In addition,the feed forward control of MPC is adopted to improve the performance of compensating the measurable disturbances,and to greatly increase the effectiveness and robustness MPC controller.Finally,the step-like disturbance and two different external forces are employed in evaluating the performances of designed MPC controller,the means square errors(MSE)of the tilt angle and rotation angle response are also calculated.Then the obtained result of the MPC controller is compared with linear quadratic regulator(LQR)controller,results shows that the performance MSE of the MPC controller will reduce more than 50%relative to the LQR controller,which can demonstrate the designed MPC controller has excellent superiority,reliability and robustness for controlling the TWIPR.

关 键 词：双轮自平衡机器人 倒立摆系统 模型预测控制 解耦单元 

分 类 号：TH16[机械工程—机械制造及自动化] TP273.3[自动化与计算机技术—检测技术与自动化装置]