Title:Second-Order Non-Convex Optimization for Constrained Fixed-Structure Static Output Feedback Controller Synthesis

Abstract:For linear time-invariant (LTI) systems, the design of an optimal controller is a commonly encountered problem in many applications. Among all the optimization approaches available, the linear quadratic regulator (LQR) methodology certainly garners much attention and interest. As is well-known, standard numerical tools in linear algebra are readily available which enable the determination of the optimal static LQR feedback gain matrix when all the system state variables are measurable. However, in various certain scenarios where some of the system state variables are not measurable, and consequent prescribed structural constraints on the controller structure arise, the optimization problem can become intractable due to the non-convexity characteristics that can then be present. In such cases, there have been some first-order methods proposed to cater to these problems, but all of these first-order optimization methods, if at all successful, are limited to only linear convergence. To speed up the convergence, a second-order approach in the matrix space is essential, with appropriate methodology to solve the linear equality constrained static output feedback (SOF) problem with a suitably defined linear quadratic cost function. Thus along this line, in this work, an efficient method is proposed in the matrix space to calculate the Hessian matrix by solving several Lyapunov equations. Then a new optimization technique is applied to deal with the indefiniteness of the Hessian matrix. Subsequently, through Newton's method with linear equality constraints, a second-order optimization algorithm is developed to effectively solve the constrained SOF LQR problem. Finally, two numerical examples are described which demonstrate the applicability and effectiveness of the proposed method.