Robust missile autopilot design using a generalized singular optimalcontrol technique

The generalized singular linear quadratic (GSLQ) control technique is developed to design an optimal time-varying trajectory tracking system. The feedforward command in this system is generally computed by integrating a reduced-order system backward in time with the Future desired trajectory and the estimated disturbance or nonlinearity as the input. The output feedback control law is designed using the GSLQ contra1 techniqqe. The feedback gain matrix is synthesized to ' minimize tracking errors with pole placement capability to satisfy the control activity requirements. The resulting tracking system is capable of determining the current optimal control strategy based on the future desired trajectory. The modeling error terms such as the uncertainty, nonlinearity, and the anticipated forcing function are included in the GSLQ control problem formulation, enabling the resulting control law to adapt to these modeling changes as long as they can be approximately estimated on-line. An application of the GSLQ technique to a bank-to-turn (BlT.) missile coordinated autopilot system design is presented. The time varying tracking autopilot is f rmulated as an optimal linear tracking system pro lem consisting of an adaptive feedforward control er and a robust output feedback controller wi if robust output feedback gains, both designed b the GSLQ control technique. The closed loop system of the resulting control law is stable for a wide range of flight conditions with little changes in the location of the closed loop eigenvalues. The control loop frequency response of six flight conditions during the terminal phase are presented to show the robustness of an output feedback controller design using the GSLQ technique. Simulations of the time responses of the tracking system with sinusoidal wave disturbances and pitch, roll, and yaw nonlinear couplings are also presented to show the GSLQ control technique for tracking autopilot with adaptive feedforward control.