气动肌肉并联关节的位姿轨迹跟踪控制

针对多输入多输出的气动肌肉并联关节，建立包含任务空间负载动态方程、容腔压力动态方程和高速开关阀平均流量方程的多阶动态系统数学模型。为保证气动肌肉并联关节系统良好动态特性的同时具有高精度的位姿轨迹跟踪，采用基于非连续投影算法的自适应鲁棒控制策略。该策略通过自适应参数估计来消除因气动肌肉并联关节系统动态数学模型的参数未知而引起的较大参数不确定，通过鲁棒反馈来消除因气动肌肉的伸缩力模型误差、摩擦力时变和关节系统的不可知干扰等引起的严重非线性不确定，且控制器基于反推设计，对多输入多输出的多阶耦合动态系统具有很好的适用性。试验结果表明：所研究的气动肌肉并联关节阶跃响应的静态误差小于0.09°，连续轨迹跟踪的标准误差小于0.15°，且具有较强的自适应性和鲁棒性。

Posture Trajectory Tracking Control of Parallel Manipulator Driven by Pneumatic Muscles

The multiple-order dynamic mathematical models of the multiple-input-multiple-output (MIMO) parallel manipulator driven by pneumatic muscles are developed, which comprise the dynamic model of parallel manipulator in task-space, the pressure dynamic model of pneumatic muscles and the average flow rate model of fast switching valves. In order to achieve highly accurate posture trajectory tracking with the guarantee of good transient performances in the parallel manipulator driven by pneumatic muscles, a discontinuous projection-based adaptive robust control strategy is proposed. The control strategy utilizes the parameter adaptation to eliminate the large parametric uncertainties, which arise from unknown parameters of dynamic mathematical models, and the robust feedback term to attenuate the rather severe nonlinear uncertainties, which arise from muscle force errors and time-varying friction forces of pneumatic muscles and unknown disturbances of parallel manipulator. Moreover, the proposed controller is based on backstepping design technology and so is fit for the MIMO multiple-order coupling dynamics system. Experimental results indicate that the parallel manipulator has strong self-adaptability and robustness with the steady error of step response being less than 0.09° and the average tracking error of continuous trajectory tracking being less than 0.15°.