FEEDFORWARD CONTROL DESIGN UNDER INPUT AND OUTPUT CONSTRAINTS
Topic: All
14 novembre 2005, Salle R05, au Centre Automatique et Systèmes, Fontainebleau
14h.30 : Knut GRAICHEN, Institut für Systemdynamik, Universität Stuttgart.
A common feedforward control problem is the realization of finite-time setpoint transitions. Typical examples are e.g. position changes of mechatronic systems or load changes in process control. In case of flat systems, the required inversion of the system is purely algebraic. For non-flat systems however, the inversion-based feedforward control design requires the numerical solution of the internal dynamics. This talk presents a new approach for the feedforward control design of nonlinear systems. The inversion-based design treats the setpoint transition as a two-point boundary value problem (BVP) in the coordinates of the input/output normal form. Thereby, a sufficient number of free parameters is provided in the output trajectory to solve the overdetermined BVP of the internal dynamics. The BVP with free parameters can be numerically solved e.g. with the Matlab function bvp4c. The approach also allows to directly incorporate constraints on the input, the output, and its time derivatives within the formulation of the BVP. The feedforward control design and the incorporation of the constraints are illustrated by swing-up and side-stepping scenarios of different double/triple pendulum systems both with nominal and experimental results.
14h.30 : Knut GRAICHEN, Institut für Systemdynamik, Universität Stuttgart.
A common feedforward control problem is the realization of finite-time setpoint transitions. Typical examples are e.g. position changes of mechatronic systems or load changes in process control. In case of flat systems, the required inversion of the system is purely algebraic. For non-flat systems however, the inversion-based feedforward control design requires the numerical solution of the internal dynamics. This talk presents a new approach for the feedforward control design of nonlinear systems. The inversion-based design treats the setpoint transition as a two-point boundary value problem (BVP) in the coordinates of the input/output normal form. Thereby, a sufficient number of free parameters is provided in the output trajectory to solve the overdetermined BVP of the internal dynamics. The BVP with free parameters can be numerically solved e.g. with the Matlab function bvp4c. The approach also allows to directly incorporate constraints on the input, the output, and its time derivatives within the formulation of the BVP. The feedforward control design and the incorporation of the constraints are illustrated by swing-up and side-stepping scenarios of different double/triple pendulum systems both with nominal and experimental results.