ANN approach for Magnetic Levitation stabilization using gradient and Quasi Newton learning

Magnetic Levitation (Maglev) stabilization has been the area of interest for various engineering fields. Classical controllers (like PID) can handle linear plants easily but when the plants are non-linear they have difficulties to deal with. This paper presents the Neural Network controller that are nonlinear in nature and can handle the controlling of nonlinear plants. The Gradient Descent algorithm is used to minimize the error function. Further the results of NN (Neural Network) controller are improved using Conjugate Gradient learning and Quasi Newton methods. The results are presented to show better tracking behavior after applying different learning algorithms.