Linear optimal control of bilinear systems with applications to singular perturbations and weak coupling (Q1355976)

This book presents linear methods for optimal control of bilinear systems with emphasis on singularly perturbed and weakly coupled bilinear systems. The unified theme of this book is the use of reduced-order subproblems to simplify and decompose computations required for the optimal control of bilinear systems. This book consists of five chapters. Chapter 1 comprises an introduction on bilinear control systems and presents several examples of real world singularly perturbed and weakly coupled bilinear control systems. In Chapter 2 we study the optimization of singularly perturbed bilinear control systems. A sequence of linear state and costate equations is constructed, and the open-loop solution of the optimization problem is obtained in terms of the reduced-order slow and fast subsystems. The ill-defined numerical problem is completely decomposed into slow and fast time scales, leading to the reduction in the size of the required computations and allowing parallel processing of information. In addition, the near-optimal ``closed-loop'' control is obtained in the form of a linear approximate ``feedback'' control law as a linear combination of the slow and fast variables (the composite control), with the coefficient matrices calculated from two reduced-order independent time varying linear-quadratic optimal control problems. The composite control is an \(O(\varepsilon)\) close to the optimal one, where \(\varepsilon\) stands for a small singular perturbation parameter. The composite control implies an \(O(\varepsilon)\) closeness of the system trajectories to the optimal ones, and an \(O(\varepsilon)\) approximation for the performance criterion. An algorithm that produces an arbitrary order of accuracy for the closed-loop approximate ``feedback'' control is also presented in this chapter. As a case study we present the approximate linear controller design of a bilinear model of induction motor drives (Figalli et al., 1984). In Chapter 2 and 3 we use the classic approach to continuous-time deterministic singularly perturbed and weakly coupled bilinear systems following the results of (Hofer and Tibken, 1984) and (Cebuhar and Constanza, 1988). In both cases, decompositions of the original systems are achieved so that the results are obtained in terms of the reduced-order subsystems. Chapter 3 considers the optimization of the time invariant bilinear weakly coupled system with a quadratic performance criterion. It has been shown how to design the linear laws by using the linear local controllers whose composition produce the optimal linear global controller. The corresponding results have been obtained for both the open-loop and approximate ``closed-loop'' control of bilinear weakly coupled systems. The presented methodology is demonstrated on a real physical system, a bilinear-quadratic optimal control problem of a paper making machine (Ying et al., 1992). The presented work on singularly perturbed bilinear control systems is mostly based on the doctoral dissertation (Aganovic, 1993), and the one on weakly coupled bilinear systems (Aganovic, 1993; Aganovic and Gajic, 1993). In the second part of the book, (Chapter 4), the modern approach based on the successive approximation technique for optimal control of general bilinear-quadratic systems is presented. The obtained results follow closely the recent work of the authors (Aganovic, 1993; Aganovic and Gajic, 1994, 1995). Both the steady state and finite-time optimization problems are considered. The presented results produce the linear optimal control law and simplify the already existing results of (Hofer and Tibken, 1984) and (Cebuhar and Constanza, 1988). Namely, it has been shown how to replace sequences of Riccati equations by sequences of Lyapunov equations whose solutions comprise the approximate linear controller for bilinear systems. In the limit these controllers tend to the optimal ones. The proposed schemes are very efficient. It takes only a few iterations to get the convergence to the optimal controller gains. The main themes of the results obtained in Chapter 5 are the linearity of the optimal control law and the power of dynamic programming and its successive approximations technique used to get the desired results. In Chapter 5 we indicate some future research problems related to the singularly perturbed and weakly coupled bilinear systems. We propose the extension of the results presented in Chapters 2 and 3 in the spirit of results obtained in Chapter 4. Also, we indicate importance of studying singularly perturbed and weakly coupled bilinear systems in the discrete-time domain. In addition, we indicate potential research on singularly perturbed and weakly coupled bilinear stochastic systems, and related linear systems with state-dependent and/or control-dependent disturbances that are described by the bilinear stochastic modes.