Skip to Main Content
In this paper, it studies the problems of the on-line identification on the nonlinear and time-lag SISO dynamic system. It puts forward the recurrent structure to linearize the input neurons of the neural network which can describe the feasibility of the algorithm, so the neural network has the dynamic on-line identification capability. Simulation results show that the input linearization dynamic recurrent network has a strong self-adaptability and robustness. It gives a new method for SISO nonlinear dynamic system identification.
Machine Learning and Cybernetics (ICMLC), 2010 International Conference on (Volume:5 )
Date of Conference: 11-14 July 2010