Data-driven Modeling and Adaptive Predictive Anti-swing Control of Overhead Cranes

This study investigates a novel data-driven model and an adaptive predictive anti-swing control law for overhead cranes. As an alternative solution to the physics-based modeling approach, a data-driven modeling framework is formulated using the feedforward neural network and extreme learning machine, approximating the nonlinear functional mapping between the system inputs and outputs. Using the proposed data-driven modeling approach, the complete input-output behavior, including the dynamics associated with sensors and actuators, is captured from experimental data. After converting the data-driven model to a state-space form, an adaptive predictive anti-swing control law is developed using the empirical model. To compensate for the modeling discrepancy resulting from abrupt parameter variations, an online parameter adaptation law for updating the data-driven model is further developed. Thus, accurate bridge/trolley positioning and rapid swing suppression are realized in ordinary and uncertain operating conditions. The asymptotic stability of the error dynamics and the boundedness in the estimated parameters are analyzed using the Lyapunov technique. Finally, three types of experiments are performed to verify the effectiveness of the proposed modeling and control methods.