Intelligent neural network implementation for SOCI development of Li/CFx batteries

Abstract

The State Of Charge Indicator (SOCI) for the Lithium Poly Carbon Monoflouride (Li/CFx) battery has a wide range of applications. However, the dynamic environmental conditions, such as the ambient temperature, can alter the characteristic response of the battery and introduce non-linear behavior. This paper discusses the in-lab development of an Artificial Neural Network (ANN) based SOCI for the Li/CFx battery. The ANN is trained on the recorded data - voltage, current and ambient temperature, to produce a non-linear model and to accurately predict the State Of Charge (SOC) of the battery. The SOC prediction is based on the recent behavior of the battery. Preliminary experimental results using recorded datasets from the Battery Design Studio are presented for the Lithium Ion battery. The working model for the Li/CFx is currently under development. The reported results demonstrated good performance of the developed SOCI, with less than 2% average relative error on data at previously observed ambient temperatures.