Adaptive fuzzy control of the molten steel level in a strip-casting process

This paper presents an adaptive fuzzy controller with nonlinear compensation and a switching control strategy to regulate the molten steel level of a strip-casting system. The proposed controller adopts an adaptive fuzzy control structure which is reported to be superior in performance to neural-network control techniques. The adaptive fuzzy controller is robust due to its fuzzy representation of the controller, and is adaptive due to its training capability. In order to compensate for the inevitable model-estimation error, the controller introduces an additional fixed gain switching control term. Additionally, a nonlinear compensation term is used in this system to eliminate disturbances due to variations in gap and speed. Combining all these terms, the proposed controller is able to achieve zero steady-state error asymptotically, with robustness and an adaptation capability inherent in the adaptive fuzzy controller.