IFT-LPV : data-based tuning of fixed structure controllers for LPV systems

Fixed structure controllers are widely used, however the tuning thereof can be cumbersome and gives no guarantee of optimality, especially when the system is Linear Parameter-Varying (LPV). Iterative Feedback Tuning (IFT) is a technique for the optimisation of a parameterised controller based on closed-loop experiments. This paper extends the applicability of IFT to LPV systems for the case where the LPV scheduling parameters are measurable but cannot be controlled. The closed-loop LPV system matrices are factorised such that the effect of the scheduling parameter on the IFT gradient estimates can be compensated. A suffcient number of IFT experiments are performed to estimate the cost gradient and tune the parameters. The method is validated successfully via a simulation study for a special case with an LPV system