摘要
Linear quadratic Gaussian with loop transfer recovery （LQG/LTR） gain scheduling technique is utilized to design gain scheduling autopilot for surface-to-air missile. In order to eliminate the artificial uncertainties that the traditional ＂trial and error＂ de- sign process introduces into system, a method to design target loop based on pole assignment is proposed, which provides an explicit algorithm to construct the matrix differential Riccati equation （MDRE） based on the expected poles determined by the performance specifications. Meanwhile, it is proved that by introducing integrators to augment plant dynamics the fast modes of LQG/LTR gain scheduling controller can be restrained effectively, which alleviates an obstacle for the engineering application of LQG/LTR gain scheduling technique. The proposed method is applied in the design of LQG/LTR gain scheduling autopilot for a surface-to-air missile. The design and simulation results indicate that the fast modes of controller are eliminated obviously, and that the dynamic characteristics of autopilot are stable when flight Mach number and altitude vary.
Linear quadratic Gaussian with loop transfer recovery （LQG/LTR） gain scheduling technique is utilized to design gain scheduling autopilot for surface-to-air missile. In order to eliminate the artificial uncertainties that the traditional ＂trial and error＂ de- sign process introduces into system, a method to design target loop based on pole assignment is proposed, which provides an explicit algorithm to construct the matrix differential Riccati equation （MDRE） based on the expected poles determined by the performance specifications. Meanwhile, it is proved that by introducing integrators to augment plant dynamics the fast modes of LQG/LTR gain scheduling controller can be restrained effectively, which alleviates an obstacle for the engineering application of LQG/LTR gain scheduling technique. The proposed method is applied in the design of LQG/LTR gain scheduling autopilot for a surface-to-air missile. The design and simulation results indicate that the fast modes of controller are eliminated obviously, and that the dynamic characteristics of autopilot are stable when flight Mach number and altitude vary.
