Optimal filtered predictive control – a delta operator approach

Abstract

Using the framework of the discrete delta operator a new generalized predictive controller is developed. Nominal stability is assured by an end-point state weighting. Starting with a controller description in state space form, the nominal stability is proven and the minimal end-point weighting matrix is derived. The new predictive control method is analyzed using the state-space techniques and a polynomial description suitable for real-time implementation is given. An optimal filtering capability is involved by the Kalman design. A final simulation example of a flexible mechanical system underlines the improved stochastic disturbance rejection.