摘要:This paper presents an experimental validation of a nonlinear predictive control law based on quasi-Linear Parameter Varying (q-LPV) modelling. q-LPV modelling is used to schedule the plant model, thereby turning the nonlinear optimization linear but time-varying, by imposing a scheduling sequence which is found iteratively. The pendubot is an interesting benchmark problem as the loci of equilibria lie on a nonlinear manifold; furthermore, the plant becomes increasingly hard to control the more the operating condition approaches one of the two uncontrollable equilibria. Unprecedented results are obtained using this control strategy by driving the system very close to said uncontrollable pose.