Application of Neural Networks to MRAC for the Nonlinear Magnetic Levitation System

This paper investigates the application of neural networks (NNs) to conventional model reference adaptive control (MRAC) for controlling the real plant of the nonlinear magnetic levitation system. In the conventional MRAC scheme, the controller is designed to realize the plant output convergence to the reference model output based on the assumption that the plant can be linearized. This scheme is effective for controlling a linear plant with unknown parameters in the ideal case. However, it may not be assured to succeed in controlling a nonlinear plant with unknown structures in the real case. We incorporate a neural network in the MRAC to overcome this problem. The control input is given by the sum of the output of the adaptive controller and the output of the NN. The NN is used to compensate for the nonlinearity of the plant that is not taken into consideration in the conventional MRAC. We developed an efficient method for calculating the sensitivity of the plant that is utilized in the NN to perform the backpropagation algorithm very efficiently. The plant of the magnetic levitation system has inherent strong nonlinearities due to the natural properties of the magnetic fields and uncertainties. Therefore, to confirm the effectiveness of our proposed controller, we implemented our proposed controller in real time on an experimental test bed of a magnetic levitation system. Finally, experimental results verified that the proposed control strategy has the advantages of tracking desired output perfectly and reducing the error.

[1]  M. S. Ahmed.  Neural net based MRAC for a class of nonlinear plants , 2000, Neural Networks.

[2]  D. Cho,et al.  Sliding mode and classical controllers in magnetic levitation systems , 1993, IEEE Control Systems.

[3]  P. Parks,et al.  Liapunov redesign of model reference adaptive control systems , 1966 .

[4]  Chang-Chieh Hang,et al.  Comparative studies of model reference adaptive control systems , 1973 .

[5]  P. Ramadge,et al.  Discrete-time multivariable adaptive control , 1979 .

[6]  Panos J. Antsaklis,et al.  Neural networks for control systems , 1990, IEEE Trans. Neural Networks.

[7]  George W. Irwin,et al.  Direct neural model reference adaptive control , 1995 .

[8]  Kumpati S. Narendra,et al.  Identification and control of dynamical systems using neural networks , 1990, IEEE Trans. Neural Networks.

[9]  I. D. Landau,et al.  A survey of model reference adaptive techniques - Theory and applications , 1973, Autom..

[10]  Marios M. Polycarpou,et al.  Learning and convergence analysis of neural-type structured networks , 1992, IEEE Trans. Neural Networks.

[11]  Hong Wang,et al.  A direct adaptive neural-network control for unknown nonlinear systems and its application , 1998, IEEE Trans. Neural Networks.

[12]  Martin T. Hagan,et al.  An introduction to the use of neural networks in control systems , 2002 .

[13]  Kwang Y. Lee,et al.  Diagonal recurrent neural networks for dynamic systems control , 1995, IEEE Trans. Neural Networks.

[14]  Rolf Isermann,et al.  Adaptive control systems , 1991 .

[15]  Jianming Lu,et al.  A Method of Model Reference Adaptive Control for MIMO Nonlinear Systems Using Neural Networks , 2001 .

[16]  Bin Yao,et al.  Neural network adaptive robust control with application to precision motion control of linear motors , 2001 .