Decentralized self-tuning observers for processing plants with continuous moving webs

This paper presents two different decentralized self-tuning observers for the estimation of states and disturbances in processing plants with continuous moving webs, which are highly complex mechatronic systems. The main problem in the control for continuous processing plants is to control the web tension force, a very important technological parameter, but no sensors are allowed. Using only the measurable motor speed, the observers developed provide the correct estimation of the web tension force in face of the two nonlinear, time-variant disturbances modulus of elasticity and friction torque. Since the disturbances act on the same processing state, they must be identified separately. Both self-tuning observers are based on the classical Luenberger disturbance observer concept but improve this concept by intelligently utilizing the steady state information after a reference variation. Thus, the drawbacks of the classical Luenberger concept are overcome and the range of application of web tension force observers is extended to processing plants where modulus of elasticity and friction torque are unknown. Furthermore the requirements for the power electronic devices and drives due to the nonlinear control are discussed.