Identification And Control Of Nonlinear Systems Using Neural Networks

After briefly describing the state-of-the-art on dynamic modeling and motion control for the constrained flexible-link manipulators, this dissertation focuses its emphasis on dynamic modeling and intelligent control for a class of constrained flexible-link manipulators. Main work of this research consists of two independent and associated parts: Part I, including Chapter 2 and Chapter 3, studies the dynamic modeling and model simplification problem for the constrained flexible-link manipulators. Using D'Alembert-Lagrange principle and the modal expansion technique, this paper establishes a set of nonlinear ordinary differential equations describing the dynamic behaviors and vibration modes of a planar two-link manipulator with a holonomial constraint. The superiority of this dynamic model over existing dynamic models for same type of manipulator is the very-simple model configuration and the higher precision. Besides, this model is of the almost same form as that for general unconstrained rigid-link robot manipulators. By introducing a variable transformation, above-established dynamic model is further simplified into a model in the form of nonlinear equations so that some mature nonlinear system theories can be combined into the study of motion control for constrained flexible-link manipulators. Part II is composed of Chapter 4 through Chapter 6. In this part, the intelligent control problem of constrained flexible-link manipulators is comprehensively studied. Based on the dynamic model presented in Part I, some new intelligent control schemes like variable structure hybrid position/force control, adaptive fuzzy logic control, and genetic algorithm (GA)-based adaptive fuzzy logic control, etc., are proposed and applied to carry out the hybrid position/force control for a two-link constrained flexible...