Nonlinear System Identification Using Dynamic Memristor-Based Reservoir Computing System

Abstract

Nonlinear systems have attracted a lot of attention because of their widespread existence in nature and life. Among them, the modeling and prediction of nonlinear systems is the focus of the research field of nonlinear systems. Although the traditional neural network has achieved good results, it is not conducive to being applied to practical problems due to the unsatisfactory training speed and large energy consumption. In this paper, considering the nonlinear characteristics of the memristor and the fast training speed of reservoir computing, we combine memristor and reservoir computing. Lorenz time series prediction and second-order nonlinear system modeling tasks are demonstrated. The results show that our model performs well in nonlinear time series prediction and nonlinear system model identification, the feasibility of the method is demonstrated. This is of great significance to the study of nonlinear systems and can be effectively applied to the analysis of nonlinear systems.