Train Coupling Robot Based on Extended Implicit Self-Correction Generalized Predictive Control

We propose A controller based on generalized predictive control is used to control the robot, so as to realize the precise tracking and control of the motion trajectory of the unhooked robot. As a nonlinear walking mechanism, the train coupler robot must ensure that it can accurately catch up with the coupler to be picked up and move in sync with the coupler after catching up. Therefore, a trajectory tracking control algorithm must be designed to control the actual operation process based on the given ideal displacement curve to ensure that the robot can operate safely and stably. Generalized predictive control has the advantages of solid robustness, broad applicability, simple parameter adjustment, strong predictive ability, and fast response speed. The Elman neural network is a dynamic feedback-type recursive neural network that combines local memory units with local feedback links, with good adaptive, self-organizing, intense learning, fault-tolerant, and anti-interference capabilities. In addressing the modeling difficulties caused by the nonlinear characteristics of the train coupler robot, a generalized predictive control algorithm is adopted for tracking control. The use of the Elman neural networks helps to address the problem of modeling nonlinear systems. Specific experimental data indicates that this algorithm can ensure the robot operates safely and smoothly, reaching the target position and velocity.