An intelligent approach for nonlinear system identification of a Li-ion battery

Abstract

An electrochemical system is a subset of mechatronic systems which includes a wide range of batteries such as nickel-cadmium, lead-acid batteries, and lithium-ion. These systems have many nonlinear behaviors and uncertainties in their operating range. Computational intelligence approaches are more accurate and quick than conventional analytical methods for estimating and analyzing nonlinear systems. In this paper, an Adaptive Neuro Fuzzy Inference System (ANFIS) model as a black box method to identify dynamic behaviors of a lithium-ion is presented. The purposed of modeling is presentation and analysis a computational intelligence approach to estimate and identify secondary cells or rechargeable batteries. In this study, a real dataset as database for exponential modeling is used. This dataset relates to a comprehensive test on li-ion batteries that was set up by researchers in NASA Ames Research Center. The presented model contains an ANFIS model using grid partition with three signals; load voltage, measured voltage and temperature of the battery. The results illustrate that the presented approach gives rather good error minimization and convergence to identify of the li-ion battery. In addition, this approach can be used to identify other storage batteries or other similar nonlinear systems.