This book presents a wide range of techniques that lead to novel strategies for effecting intelligent control of complex systems that are typically characterised by uncertainty, nonlinear dynamics, component failure, unpredictable disturbances, multi-modality and high dimensional spaces. The underlying design philosophy is based on effecting closed-loop control in the presence of plant or environmental uncertainty and complexity by utilizing various types of neural network architectures, ranging from multilayer perceptron to radical basis function and modular network models. The uncertainty and complexity are typified by unknown nonlinear functionals, and temporal or spatial multi-modality. Deterministic and stochastic conditions, as well as continuous and discrete time dynamics are taken into consideration. The presented designs are firmly rooted in the techniques of adaptive control, reconfigurable control, multiple model control, stochastic adaptive control, lyapunov stability theory and neural networks. The techniques are shown to enhance the performance of the control system in the presence of the higher levels of complexity and uncertainty associated with modern plants, which demand superior intelligence and autonomy from the controller. The presented designs are supported both by theory and by numerous results from simulation experiments. The book also includes extensive reviews on general aspects concerning the fields of intelligent, nonlinear and stochastic control.

Unique in its systematic approach to stochastic systems, this book presents a wide range of techniques that lead to novel strategies for effecting intelligent control of complex systems that are typically characterised by uncertainty, nonlinear dynamics, component failure, unpredictable disturbances, multi-modality and high dimensional spaces.