Neural Network Compensation Control for Model Uncertainty of Flexible Space Manipulator Based on Hybrid Trajectory

A flexible space manipulator is needed to accomplish various tasks in the space in future.However, model uncertainty and the flexibility of link decrease the trajectory tracking accuracy of a space manipulator.To avoid the influences of model uncertainty and the vibration of flexible manipulator on control accuracy, the trajectory tracking and flexible suppression problems of free-floating flexible space manipulator with an attitude-controlled base were discussed in this work.First, the flexible link was viewed as an Euler-Bernoulli beam.The horizontal deformation of the flexible link was described by the assumed mode method, and a dynamic equation of the flexible space manipulator system was set up by the Lagrange method of the second kind.Second, by considering the model uncertainty of the system, the advantages of the method on the basis of model control were developed and model uncertainty was compensated for by a neural network.Moreover, a neural network compensation control scheme for rigid motion was proposed to achieve coordinated motion between the base's attitude and the joint's angles of flexible space manipulator.Subsequently, a hybrid trajectory that reflects flexible modal and rigid motion simultaneously was designed by using the concept of virtual control force to inhibit flexible vibration actively.A control scheme in line with a hybrid trajectory was proposed.Research results demonstrate that the proposed control scheme can help the base's attitude and two joints' angles of the space manipulator to track the given desired trajectories effectively and also eliminate vibration of flexible links.This study provides an effective scheme to increase control accuracy in the trajectory tracking of the space manipulator under model uncertainty and the vibration of the flexible links in practical conditions.