Optimal worst case estimation for LPV-FIR models with bounded errors

In this paper discrete time linear parameter varying (LPV) models with finite impulse response (FIR) dynamic structure are considered. Measurement errors are assumed to be bounded. In such condition optimal input sequences minimizing the worst case parameter uncertainties are derived. The main result of this paper consists in finding optimal worst case input sequences for LPV–FIR models. These are obtained suitably combining the optimal design results for standing alone invariant FIR models and standing alone nonlinear memoryless blocks. The quite relevant improvement obtained using optimal input sequences instead of random sequences is shown by simulation.