Comparison between adaptive linear controller and radial basis function neurocontroller with real time implementation on magnetic levitation system

This paper presents the real time control of Magnetic Levitation System using feedback linearization and online learning using Radial Basis function neural networks. It starts with the comparison between adaptive linear controller and neurocontroller used here. Both the controllers uses feedback linearization method and Lyapunov's stability theory to derive update laws to control the nonlinear system. Simulation results justify the use of neurocontroller over adaptive linear controller. Magnetic Levitation System which is open loop unstable is stabilized using feedback linearization approach which relies on canceling the nonlinearities, whereas the online learning is implemented via function approximation in which the radial basis function neural network is used as a function approximator. The combination of feedback linearizing control law and adaptive laws ensures that the closed-loop system is stable.