带有死区和摩擦补偿的机械臂伺服系统自适应神经网络控制

摘要

针对带有输入死区和未知摩擦的机械臂伺服系统，本文提出了一种基于神经网络的自适应轨迹跟踪控制策略。首先，建立死区的逆模型，解决死区的非线性输入问题。其次，构建摩擦力动态模型，并采用径向基核神经网络来逼近系统中的不确定项。然后，通过反演法和一阶动态面，递归设计控制虚拟量和控制器，以保证系统输出能快速跟踪期望信号，提高跟踪误差的收敛性能。最后，仿真结果验证了该方法的有效性。%In this paper, an adaptive control scheme based on the neural network is proposed for robotic manipulators with input dead zone and unknown friction. Firstly, the inverse model of dead zone is established to overcome the issue of input nonlinearity. Secondly, the friction behavior is described by a nonlinear dynamical model, and a radial basis function neural network (RBFNN) is employed to approximate the uncertainties in the system. Then, virtual control variables and the controller are designed by combing the backstepping technique and the dynamic surface control in each step to guarantee that the system output can rapidly track the desired signal and the convergence performance of track error can be improved. Finally, simulation results are given to verify the effectiveness of the proposed scheme.