ABSTRACT
This chapter introduces a family of neural network (NN) control architectures known as adaptive critic controllers as a natural extension of simpler architectures. It discusses the merits of each architecture and their shortcomings exposed, which in turn becomes the motivation for the next. The controller is then improved by the addition of a second NN capable of generating online a map of the plant’s dynamics; however, the training algorithm remains fundamentally the same. The chapter provides the readers with basic motivation, background, and description of different adaptive critic controllers by presenting a series of NN adaptive controller architectures ranging from a single NN adaptive controller to globalized dual heuristic programming. Active Fault tolerant control systems compensate for the effects of a fault either by selecting a new precomputed control law or by synthesizing a new control law online. Model predictive control, feedback linearization, and sliding mode are examples of such methods.
