Terminal sliding mode control for rotary cranes based on Extreme Learning Machine

A terminal sliding mode (TSM) control strategy based on Extreme Learning Machine (ELM) is applied for the tracking control and load sway control of a rotary crane with uncertain dynamics. The disturbance observer is designed to compensate the unknown disturbance. Then the complex crane system can be divided into two linear models. The proposed control effectively estimates the lumped uncertainty by a neural network with ELM. For the problems of slow convergence and chattering, a terminal sliding mode surface is introduced to ensure the tracking error converge in limited time, and the continuous function is used to replace the sign function to eliminate chattering. Finally, the closed-loop stability and finite-time convergence are proved. The simulation shows that the method is robust to the changes of load mass and rope length of the crane system.