Multivariate controllers for LQG self-tuning applications with coloured measurement noise and dynamic cost weighting

Abstract

A solution of the generalized LQG stochastic optimal multi variable control problem is obtained using a polynomial-matrix-systems approach. The generalized problem involves a dual performance index including the usual error and control weighting terms and also sensitivity and complementary sensitivity weighting matrices. The system model includes input disturbance, reference and coloured measurement noise signals. The cost-function weighting operators can also represent dynamical systems. The problem description was motivated by the need to construct multi-variable explicit self-tuning controllers for marine applications.