Stochastic adaptive control results and simulations (Q581307)

The book is a summary of theoretical results from the literature concerning adaptive control of stochastic systems. Two classes of systems are considered. First, systems with unknown but constant parameters. Secondly, the parameters of the system are modeled as finite state homogeneous Markov chains. The theoretical results are verified through numerous numerical simulations. The conclusion is that a modified least squares estimation method gives the best performance. The comparisons are, however, only made on ``academic'' algorithms, i.e. there are no safety nets around the controller to prevent large control signals nor checking the residuals in the estimator. The advantage of the book is the summary of algorithms, conditions and properties. The results are, however, of limited practical interest.