Dynamic feature space modelling, filtering and self-tuning control of stochastic systems. A systems approach with economic and social applications (Q1060166)

This monograph aims at unifying the design methodologies of engineering and social sciences by building the interrelationships between the concepts and methods of systems, estimation, and control theories which are being extensively used in the design of engineering systems and the quantitative tools of economics and social sciences. Although, most of the work is allotted to explaining engineering design methods, also contained in this work are interesting application examples chosen mostly from economics. After an introductory chapter, in chapter 2, the rudiments of systems theory are presented, which include dynamic system models, controllability, observability, realization theory, canonical forms and stability. In chapter 3, the ''linear structural feature space (LSF)'' model and a simpler form of it called the (LRF) model are introduced. Feature vector is defined as the ''characteristic part of the models'' and is a generalization of the state vector idea. The state space models, models used in factor analysis, etc. are shown to be special cases of the newly defined models. The next chapter is devoted to the optimal reconstruction of the feature vector, namely Kalman filtering. The derivation of the filter equations is given and the asymptotic properties of the filter are examined. Chapter 5 is on the parameter identifiability problem, the prediction error methods and their applicability to (LSF) and (LRF) models. Chapter 6 is an applications chapter in which the author tries to give the reader a taste of the social and economic applications of feature space modelling. In chapter 7, the linear-quadratic-gaussian problem formulation, asymptotic properties of its solution and its application to (LRF) models are presented. Short treatments of minimum variance control, closed-loop identifiability and self-tuning control are given. Then, self-tuning control is applied to an example problem involving a simplified model of U.S. economy. The results of the simulation experiments are illustrated and it is concluded that the adaptive control is not so successful because the initial assumptions are somewhat unrealistic and so not exactly met in this specific situation. Also, in the appendix at the end of the book, the solution methods for the generalized Lyapunov equation and the algebraic Riccati equation are discussed. This monograph may serve the purpose of a handy reference book for engineers, mathematicians, economists, and researchers in social sciences alike, and it certainly is a positive step towards closing the gap between systems and control theories and economic and social science applications.