A numerical technique for optimal output feedback using Pade approximations

Abstract

Any linear control system, including those with complex structure and dynamic compensation can systematically be formulated as a static output feedback problem where the control system parameters are contained in the output feedback matrix. For this reason, many numerical techniques have been developed specifically to solve the 7/2 optimal ouput feedback problem posed in static output feedback form. Current numerical techniques for computing the output feedback gain matrix recast the problem as a coupled set of Lyapunov equations. Various numerical techniques effectively solve the Lyapunov equations differently. Here, we propose a new numerical method for solving the output feedback gain matrix by direct solution of the cost function using Fade approximation. Using this technique, analytic expressions for cost function derivatives can be written compactly, allowing for efficient solution using second order optimiztion methods. Using a flight control system example, the new method is compared to existing optimal output feedback algorithms.