Rotation angle vibration suppression for variable-length flexible manipulator based on neural network identification with sliding-mode controller

The variable-length flexible manipulator has an impact on the flexibility of joints and the flexibility load, in which vibration at the rotation angle can occur. To suppress the vibration, a control law combining neural network identification, sliding mode control and angle-independent method is proposed. To begin with, considering friction torque and two-dimensional deformation, the variable-length flexible manipulator dynamic equations are established. In addition, an adaptive law for the neural network weight coefficients is devised through the Lyapunov stability theorem. Eventually, the simulated analysis and controlled experiments are performed. The experimental finding demonstrates this: in general, the influence of nonlinear terms on deformation is negligible. However, when the bending stiffness is low, the influence of the nonlinear term cannot be ignored. The control accuracy of rotation angle can be improved by applying neural network compensation for the uncertain part. The flexible load vibration can be suppressed by the angle-independent method. The vibration of the flexible load can be suppressed by the angle-independent method. The combination control strategy decreased the absolute error mean by 15.85%, reduced the variance of the error by 37.13%, lowered the standard deviation of the error by 20.43%, and reduced the mean acceleration by 9.21%.