Digital control of a magnetic levitation system through xPC real-time operating system: Classical, feedforward and artificial neural network.

摘要

This thesis is concerned with the development of a modular easily expandable real-time digital control system for engineering education in dynamic system modeling and controls. A magnetic levitation device, which is nonlinear, open-loop unstable, time-varying, was chosen for implementation. Host-target real time solution is implemented by using xPC Target. A classical lead-lag controller was implemented using linearization techniques for nonlinear systems. Advanced control feedforward and artificial neural network (ANN) techniques were also implemented to improve tracking performance of the system. An on-line learning algorithm, back-propagation, is implemented for the ANN by using MATLAB/Simulink programming environment. Experimental results are included in a comparative study to demonstrate the effectiveness of the various control algorithms implemented.