Learning identification of time varying parameters in nonlinear systems with initial state learning

For a class of nonlinear systems with unknown time varying parameters, a iterative learning identification method with initial state learning is proposed. The method uses the operator theory to prove that the output of identification can track the expected trajectory completely after the iterative learn of system under the arbitrary initial state, and provides the sufficient convergent condition by the spectral radius form of the method. This method not only can realize the complete identification of nonlinear systems' unknown time varying parameters in finite time horizon, but also can solve the problem that the iterative learning identification needs the rigid repetition of initial state. Simulation results verify the validity of the proposed method.