By significantly extending digital signal processing theory to include such feedback effects, this book opens new techniques to designers for implementing effective control compensators.
Control theorists have developed many elegant algorithms for the design of discrete-time compensators, but the mathematical formalism has not been readily applicable to implement them. At the same time, digital signal processing theorists have achieved results dealing with filter structures, finite precision, and pipelining which would be useful in this implementation if the results included the effects that arise due to external feedback, which is fundamental to control systems. By significantly extending digital signal processing theory to include such feedback effects, this book opens new techniques to designers for implementing effective control compensators. In order to develop specific results for digital compensator implementation, the book's investigation is limited to a single class of control problems, those associated with the steady-state, linear-quadratic-Gaussian (LQG) control framework. Controllers based on this framework have desirable performance properties in terms of robustness, multivariate formulation, and optimal nature, and are being increasingly used in real systems. After an introduction that defines the parameters of the field and a chapter on the LQG problem, the book takes up compensator structures; architectural issues (serialism, parallelism, and pipelining); finite word-length effects as they relate to quantization noise, quantitizing the coefficients, and limit cycles; and the optimization of structures. A final summary chapter presents conclusions and indicates directions for future research efforts.
The book is the fourth in the MIT Press Signal Processing, Optimization, and Control Series.
