Abstract: To avoid the control failure caused by system parameter uncertainty, we analyze the performance index of a reduced controller mixed with parts of the fast modes. We use a sixth-order roll autopilot system as an example to illustrate the timescale characteristic change caused by system parameter uncertainty. Then, we simulate and analyze the states needed to design a reduced controller and the highest order of the controller. The extended reduced controller using all potential slow modes is designed, and its control performance is verified through simulation. Results show that the control performance of the extended reduced controller is equivalent to that of the traditional reduced controller, and the proposed controller shows good adaptability for a system with uncertainties.