An extended ADALINE neural network trained by Levenberg-Marquardt method for system identification of linear systems

This paper presents a sliding-window version of online identification method for linear time varying systems based on the ADaptive LINear Element - ADALINE (Widrow and Lehr, 1990) neural network trained with Levenberg-Marquardt method which offers faster tracking of system parameter change. It is well known ADALINE is slow in convergence which is not appropriate for online application and identification of time varying system. To speed up convergence of learning and thus increase the capability of tracking time varying system parameters, our previous work added a momentum term to the weight adjustment. While the momentum does speed up convergence, it also shows overshooting or oscillating and also tracks noise closely. The Levenberg-Marquardt method is explored in this paper. Simulation results show that the proposed method provides indeed fast yet smoother convergence and better tracking of time varying parameters.