基于神经网络柔性机器人反演控制器的设计

Design of backstepping controller for flexible robot based on neural network

建立柔性机器人系统的动力学模型,对实际情况作了相应的假设,借助于神经网络函数的逼近技术,提出一种柔性关节空间机器人的自适应反演控制方案,给出基本反演控制器和神经网络反演控制器的数学模型。仿真结果表明,能实现系统关节运动控制目标,可使系统各柔性关节的振动较小,有效地控制机器人完成所期望的关节运动,对无精确模型信息的机器人实现控制,保证系统的稳定性和收敛性,满足机器人系统的控制要求,所设计的控制器在具体的机器人控制中是可行的。

The dynamic model of flexible robot system,the actual situation of the corresponding hypothesis,based on neural network function approximation technique,an adaptive inversion of a flexible joint space robot control scheme,gives the mathematical model of basic inversion controller and neural network inversion controller.The simulation results show that the control system can achieve the goal of joint movement,the smaller the vibration of flexible joint system,joint movement effectively control the robot to complete the desired,to achieve control of the non accurate model of robot,ensure the stability and convergence of the control system,to meet the demands of robot system,the controller designed is feasible in the robot control.