Experimental Implementation of Neural Control Architectures for a Class of Nonlinear Dynamic Systems

A novel neural network controller architecture is explored for the speed control of a DC motor-driven four bar linkage system which represents a class of time-varying nonlinear systems. Simulation and experimental implementation show the viability and effectiveness of this promising tool for nonlinear systems. Identification and control, as well as real-time implementation issues are considered. Comparative studies with two other neural network architectures are also reported. Importantly, they also perform well in real-time control studies for the system considered.