Clutch Mechanical Leg Neural Network Adaptive Robust Control of Shift Process for Driving Robot with Clutch Transmission Torque Compensation

Abstract

In order to improve the motion accuracy and robustness of the clutch mechanical leg for driving robot (CMLDR), and to improve the shift performance of the driving robot vehicle (DRV) for the engagement operation of CMLDR, a neural network adaptive robust control method of shift process for CMLDR with clutch transmission torque compensation is proposed. Firstly, the dynamics model of the driving robot is established considering an external interference force and dynamic modeling error of the mechanical leg. Secondly, the upper clutch controller including a two-parameter shift module, a switch controller and a finite-time linear quadratic regulator (LQR) of the clutch transmission torque is designed. Thirdly, the estimator of clutch engagement torque through a Kalman filter in the process of shift is constructed. Then, the neural network adaptive robust (NNAR) controller of the mechanical leg is designed. Finally, the proof of stability analysis for NNAR controller is conducted. Experiment and simulation results show that the designed controller of the mechanical leg has a strong anti-interference ability. And the DRV accurately track the target speed after compensating for the transmission torque.