Closed-loop model validation; an application to an unstable experimental system

Abstract

Presents a model validation technique based on closed-loop, sampled, experimental data, which is applicable to open-loop unstable systems. Closed-loop probing signals are used to generate the I/O information for the stabilized system. Sampled measurements are then taken to check the validity of the open-loop unstable system model. A general closed-loop model validation test is derived by applying the lifting theory on the hybrid system at the sampling period. Sufficient conditions are also derived using decimated data sets to alleviate the computation problem. Application to an inverted pendulum experimental system is illustrated.