Abstract: This paper introduces an augmented LQR method.This method uses the derivatives of the original system states and command tracking errors as augmented states.By virtue of this,the augmented LQR method results in a robust multivariable system which can eliminate steady state errors,keeping as good robustness as LQR.When all system states are not measurable,a reduced order state observer is introduced,which is designated as near optimal LQR method.The resulted control system has superior performance compared to LQG,and is able to meet the requirements for aeroengine control system.Finally,simulation results attests to the properties of these methods.