Design of LOG/LTR controller: application to a turbofan engine

The approach of G. Stein and M. Athans (1987) for the solution of a specific H₂ optimization problem by LQG/LTR (linear quadratic Gaussian/loop transfer recovery) is revisited. A colored measurement noise whose spectrum only depends on the weighting filter of the H₂ cost function is introduced in the plant model. A LQG/LTR controller is then designed for this augmented plant model and it is shown that this controller solves the H ₂ optimization problem for the original plant model. The proposed methodology is then applied to design a LQG/LTR controller for the turbofan engine considered by S. Garg (1989). Contrary to the method of Garg, no plant scaling is necessary to meet the design specifications. Good stability margins are obtained at the physical interface of the plant, while this could not be achieved in the work of Garg