【作者】 张祥；

【导师】 黄田； 杨志永；

【作者基本信息】 天津大学 ， 机械制造及其自动化， 2005， 硕士

【摘要】 本文紧密结合国家自然科学基金项目,以二自由度高速轻型并联Diamond机械手为对象,研究提高其速度与精度的控制策略和方法,并取得如下成果:借助矢量法构造机构的运动学逆解模型,消元法导出机构正解模型。在此基础上,利用虚功原理建立其刚体动力学模型,从而为轨迹规划与机械手控制提供必要理论前提和技术基础。提出了一种适用于高速轻型并联机械手的轨迹规划方法。该方法以机械手最小运行周期为目标函数,同时兼顾其对伺服系统输入力矩的影响,在操作空间进行轨迹规划,并通过运动学逆解模型将其映射到关节空间进行正弦、多项式和修正梯形模式选优。在此基础上,设计了一柔性有障碍弧线过渡路径规划方案,从而为机械手高速、高精度控制奠定了坚实的基础。提出了一种新的定常PID参数整定方法。该方法综合考虑机电耦合效应和全域性能指标,在操作空间设计一带权重轨迹,并结合运动学与逆动力学模型,建立以末端执行器均方根误差为目标函数的整定方案,从而有效的解决了传统整定方法因其多目标函数相互制约而导致最优解很难得到的问题,保障了机械手在全域空间性能趋优。提出了一种基于BP神经网络的自适应控制策略,深入探讨了其在高速并联机构控制方面实时性与可靠性问题,其4-5-3型网络时变PID算法有效的解决了恒定增益难以保证高速并联机械手这类高度非线性的时变系统动态高品质的问题,从而极大地提高了系统控制精度。提出了一种外环PID控制与内部非线性反馈环节相结合的计算力矩控制策略。该策略通过构建非线性反馈质量矩阵有效的将机构角加速度与系统输入力矩联系起来,从而有效地补偿了动力学因素引起的跟踪误差,确保了高速并联机械手高精度作业。以上工作对于推动高速轻型并联机械手控制方法研究,提高我国在自动化行业的竞争实力将具有非常深远的意义。更多还原

【Abstract】 Aimed at a two DOF light-weight parallel manipulator with high speed, severalcontrol strategies to enhance its speed and precise are presented. The following workshave been accomplished.With the aid of vector method, the inverse kinematic model has been developed.And the inverse dynamic model has also been formulated by the means of virtualwork principle. This provides necessary theory foundation for the path planningand control strategy research.An effective path planning method for high-speed parallel manipulator has beenproposed, taking the minimum pick-and-place time as the objective and at thesame time considering its affection on servo control system. The path planning iscarried out in the working space firstly and then through inverse kinematic modeltransformed into joint space to contrast sine, polynomial and trapezoid patterns.Based on this, a flexible arc trajectory planning method is presented, whichmakes the higher speed and higher precise possible.A new method on parameter tuning for the fixed-parameter PID controller hasbeen presented. Taking the elect-mechanical coupling and the whole workingspace performance into account, together with the aid of the kinematic model,dynamic model and the designed paths with coefficients, the tuning scheme isbuilt up, which takes the mean square root error as the objective. It solves theproblem of optimum results difficult to get in the traditional way since it takesthe interactional indexes as objective. As a result, the high performance of themanipulator is successfully guaranteed.A self-adaptive control strategy based on BP neural network has been broughtforward. The real-time capability and reliability of the means applying the neuralnetwork into the area of has been deeply studied and a 4-5-3 structure real-timevaried PID arithmetic has been also carefully designed. Thus the problem ofhigh-quality performance difficult to realize in fixed-parameter control isovercome.A computed torque method with the framework of outer PID control plus innernonlinear feedback has been designed. The following error result from dynamicfactors is compensated through a mass matrix relating the joint acceleration withthe actuator torque. Hence, the high quality control performance has beenachieved.The works accomplished in this thesis are very meaningful for promoting thecontrol strategy research on high-speed and light-weight parallel manipulators andimproving our capacity in the competitive automatic industries.更多还原