Identification and stochastic adaptive control (Q1189435)

The objects study in this book are stochastic systems, essentially in discrete time at a post-graduate level. The basic model used to describe such systems are ARMAX models, and estimates are extended least squares and stochastic gradient. Basic results on martingales and on linear stochastic systems are given in the first three chapters. From chapter 4, sharp results and their complete proofs, founded on very recent works --- in particular on a large number of nice publications of the authors --- are given and presented in a nice way. In chapter 4, consistency of coefficients estimate is proved under a weakened excitation condition. Chapter 5 gives an analysis for the convergence and optimality of adaptive trackers. A diminishing dither (chapter 6) added to the control makes the parameter estimate, in an adaptive control, consistent. Chapter 7 discusses estimating the order and the time delay for stochastic feedback control systems. Keeping the consistency of parameter estimates, chapter 8 designs the optimal adaptive control minimizing the tracking error or the general quadratic loss function. Chapter 9 is concerned with the approximation of a system of infinite order by a finite order one. Chapter 10 and 11 are devoted to the identification and adaptive control of stochastic systems with time varying parameters. Finally, chapter 12 is concerned with continuous time stochastic systems. My conclusion: A capital book to stimulate research in systems and control sciences, statistics, applied probability and econometrics.