Cycle Life Prediction of Battery-Supercapacitor Hybrids Using Artificial Neural Networks

Abstract

The cycle life of batteries and battery-supercapacitor hybrid systems was predicted using artificial neural networks. The presented techniques are able to predict the cycle life of a device based on a short (around 4% of the average cycle life) initial segment of the discharge curve. The prediction showed good performance with a correlation coefficient above 0.95. We were able to improve the predication further by considering ready measurements about the device and usage.