![]() | Only 14 pages are availabe for public view |
Abstract In the last decade, non-linear semi-denite programming problem (NSDP) have attracted the attention of many authors in the optimization community. For instance, Jarre [11] introduced an interior point method for non-convex semide nite programs. Leibfritz and Mostafa [14] proposed an interior-point trustregion method for a special class of NSDP problems resulting from the continuoustime static output feedback (SOF) problem. Kocvara and stingl [12] considered an augmented Lagrangian method for a similar NSDP problem. Sun et al. [30] investigated the rate of convergence of the augmented Lagrangian approach for NSDP. Correa and Ramirez [5] proposed a sequential semi-denite programming method for non-linear semi-denite programming problems. Yamashita and Yabe [34] studied local and super-linear convergence of a primal-dual interior point method. Mostafa et al. [25] introduced an SQP method globalized using line search for solving certain NSDP problem originating from discrete-time output feedback control systems. He introduced an augmented Lagrangian SQP method for two NSDP originating from such systems. Recently, non-linear optimization techniques have been employed by several researchers to solve NSDPs arising in optimal control (see for example, [21], [25]). Feedback and control is a central feature of life. The process of feedback governs how we grow, respond to stress and challenge, and regulate factors such as body temperature, blood pressure and cholesterol level. The mechanisms operate at every level, from the interaction of proteins in cells to the interaction of organisms in complex ecologies. Feedback and control are almost everywhere. One can virtually link the powerful word ”control” to almost anything, such as ”diet control” ,”nancial control”, ”pest control”, ”motor control”, ”robot control”, etc. One can also say that ”power is nothing without control,” which is believed to be correct in both social and technological contexts. Feedback is an intuitive means for control. For example, when you feel cold (sensing), you add one more layer of cloth (decision and then control action) to keep yourself warm and comfortable (objective). This is biological feedback due to a change in the environment. The CHAPTER 1. INTRODUCTION 4 early development of automatic control devices can be traced back to the ancient water clock in Alexandria, Egypt, or to the ancient compass vehicle developed about 2,000 years ago during the Han Dynasty in China. The ying ball governor invented by James Watt in 1788 is regarded as the rst widely used automatic feedback control system. Theoretical research on control systems was initiated by the study of stability problems involving dierential equations pioneered by the work of Maxwell in 1868, Routh in 1874, and Hurwitz in 1895. Control strategy design problems were rst proposed by Minorsky in 1922. Robust control is a very attractive new area in control systems design. Modern robust control investigations were started by Zames in 1981, where optimal control problems were formulated as the minimization of norms in Hardy spaces. The state space solution to such problems by Doyle et al. in 1989 is a signicant computational contribution. Most feedback system analysis and design tasks can be solved easily using a computer. Therefore, suitable computer software is essential for control system investigations. Control systems can be broadly broken up into two major categories: open loop control and closed loop control. Systems in which the output quantity has no eect upon the process input quantity are called open-loop control systems. On the other hand, closed-loop control systems are those in which the output has an eect upon the process input quantity in such a manner that the desired output value is maintained (see e.g., [8]). The design of an optimal output feedback controller that meets a desirable performance criterion is an important non-linear and non-convex problem in optimal control theory (see e.g., [22], [20]). Over the last two decades, several problems in system and control theory were reduced to some special class of non-linear semide nite programs. Although the NSDP formulations of control systems were made popular in the mid 1990s, computational methods for solving general non-convex NSDPs were not yet developed. The static output feedback problem is one of the most important open problems in optimal control theory (see [2] and [3]). The optimal SOF control problem was solved for the discrete-time case using unconstrained optimization techniques (see<br. |