Nonlinear Adaptive Vibration Suppression Control for Compound Motion of a Rigid-Flexible Coupling Hoisting Robot Based on the Neural Network

The vibration suppression of the rigid-flexible coupling hoisting robot (RFCHR), which exhibits nonlinear and underactuated characteristics, remains a significant challenge in engineering. Most existing vibration suppression controllers overlook underactuated variables or rely on indirect energy-based strategies. Even in methods that consider underactuated variables, gain parameters often fail to adapt to time-varying system signals and disturbances, limiting the performance of the controller. To address this, a nonlinear adaptive control method is proposed that incorporates the relevant variables of the underactuated payload swing angles into the controller design, enabling faster and more accurate suppression of vibrations during compound motion. Furthermore, a grey wolf optimization algorithm-based backpropagation neural network (GWO-BPNN) is designed to perform online tuning of the controller's gain parameters. The learnable parameters are optimized by the algorithm, thereby enhancing the adaptability and generalization of network. Finally, the superiority and robustness of the method are verified through experiments on the experimental platform.