挖掘机器人阀控缸系统RBF神经网络参数辨识

为提高液压挖掘机器人工作装置轨迹规划控制精度,减小按照理想模型进行控制的阀控缸系统存在的控制误差,获得更接近实际状况的阀控缸系统控制模型,采用RBF神经网络方法,建立含阀控缸系统待辨识参数及Jacobian信息的线性方程组.以挖掘机斗杆油缸为研究对象,经实验获得油缸进回油压力、斗杆倾角参数,辨识出阀控缸模型中阀的增益系数kq、体积模量Eoil和内泄漏系数Cli.最后通过对阀控缸系统进行力控制实验对比研究,验证了采用辨识参数的系统模型控制精度较好,有很强的鲁棒性.

RBF Parameter Identification of Valve-Controlled Cylinder System for Excavator Robot

To improve the planned control accuracy of trajectory of the working parts of a hydraulic excavator robot, i.e., to develop an applicable control model for the valve-controlled cylinder system instead of the ideal one so as to make the control of the robot more actual with control error decreased. By way of RBF neural network, a set of nonlinear equations was deduced involving the parameters to be identified for the valve-controlled cylinder system and relevant Jacobian information. With the bucket arm/cylinder system of the excavator robot investigated via tests, such parameters of cylinder's oil inlet/outlet pressure and rake angle of bucket arm were given to identify the valve's gain coefficient kq, oil volumetric coefficient Eoil and inner leak coefficient Cliin the valve-controlled cylinder model. It is verified that the model involving identification parameters has high control accuracy and robustness as the result of a comparative force control test for the valve-controlled cylinder system.