Data-driven Adaptive Optimization Recursive Identification for a Discrete-time Nonlinear System

In this paper a novel data-driven adaptive optimization recursive identification (DD-AORI) is proposed which applies for discrete-time nonlinear systems. The dynamic linearization technique is used to equivalently transform discrete-time nonlinear systems into the form of data-driven model. An optimal and convergent estimation for quickly time-varying parameters including its parameter update rate and covariance update rate is obtained by optimization indexes and Lyapunov convergence. The method improves the tracking ability of time-varying parameters by increasing the use of historical data. The damping factor is used to improve the dynamic effect of the system. To improve the estimation effect, the utilization ratio of new data is increased by forgetting factor. Simulation results are presented to verify the effectiveness of the proposed method.