节点文献
重载大惯性液压驱动系统的神经网络近似内模控制
【作者】 刘振；
【导师】 邓华；
【作者基本信息】 中南大学 ， 机械电子工程， 2010， 硕士
【摘要】 液压驱动系统由于具有强非线性,参数不确定性等特点而很难对其进行精确运行控制,其中主要的难点是建立系统的精确数学模型。本文以250T锻造操作机夹钳旋转机构液压驱动系统为对象,研究了神经网络近似内模控制在重载大惯性液压驱动系统控制中的应用。采用神经网络非线性辨识的方法对重载大惯性液压驱动系统进行辨识,建立了系统的神经网络模型,避免复杂非线性不确定性液压系统难以建立精确数学解析模型的问题。考虑系统的强非线性,模型不确定性和工作点的变化,将神经网络近似内模控制应用于重载大惯性液压驱动系统。该控制器中,系统的神经网络逆控制器可以直接由辨识得到的系统神经网络模型导出,不需要训练第二个神经网络。仿真结果表明,与PID控制器相比,神经网络近似内模控制器能较好地抑制系统模型不确定性和工作点变化的影响。考虑液压马达内部动态摩擦特性对液压驱动系统动态性能的影响,研究了基于LuGre模型的重载大惯性液压驱动系统的摩擦补偿控制,其中LuGre模型中的各参数通过转速——摩擦力矩实验辨识估计得到。由于重载情况下可能需要多马达共同驱动并且存在同步误差,分别研究了基于速度反馈和压差反馈的交叉耦合同步误差补偿控制,仿真结果表明,基于压差反馈的交叉耦合同步误差补偿控制不仅能实现马达输出转速的同步,同时也能使马达输出口压力值保持一致,实现负载均衡。
【Abstract】 Due to high nonlinearity and parameter uncertainties, the precious motion control of hydraulic systems is not easy。One of the difficulties is to build their exact mathematical model. This thesis focuses on the application of approximate internal model-based neural control (AIMNC) strategy in heavy-load large-inertia hydraulic systems for the grippers of 250 Ton forging manipulators.A heavy-load large-inertia hydraulic system is identified using nonlinear identification method based on neural networks and the neural network model of the system is thus built, which solves the problem of building an exact mathematical model of complex, uncertain and nonlinear hydraulic systems.In view of the high nonlinearity, model uncertainties and variation of operation points of the system, the application of a novel AIMNC strategy in heavy-load large-inertia hydraulic systems is studied. The neural inverse control law can be derived directly from the identified neural network model without further training and only one neural network needs to be trained. Simulation studies demonstrate that the AIMNC strategy exhibits better control performance and robustness to model uncertainties and variation of operation points than the PID controller.In view of the influence of internal dynamic friction on hydraulic systems, the friction compensation control of heavy-load large-inertia hydraulic systems based on LuGre model is studied. The parameters in the LuGre model are estimated through a velocity—torque experiment.As two hydraulic motors are needed to drive heavy-load large-inertia hydraulic systems and the synchronization error among hydraulic motors, a cross-coupled controller for synchronization error compensation based on feedback of velocity and pressure is studied for two hydraulic motor driving systems. Simulation studies demonstrate that the cross-coupled controller based on pressure feedback can achieve better velocity synchronization and load balance.