Session: 10-01 Interactive Presentations

Paper Number: 99241

99241 - Estimation of State of Health Degradation of Thin Flexible Li-Ion Batteries Subjected to Accelerated Life Cycling With Randomized Levels of Charge-Discharge and Varying C-Rates 

The increasing need for wearable electronics, fitness accessories and biomedical equipment has prompted an increase in thin flexible battery research and development. The degradation of such power sources used in consumer electronics devices is important from a number of perspectives: it allows the manufacturer to determine the device warranty, it influences the user's purchasing decision, and it can also be used to inform the user of their device's battery health and remaining useful life in real-time. Batteries are often submitted to accelerated life cycling experiments with diverse operating circumstances to attain these goals via empirical approaches, and the degradation data acquired is then utilized to simulate their SOH degradation. For the task of model construction, many modeling techniques like as simple regression and complicated neural networks are utilized to estimate the battery SOH. These models are often built with training data derived from battery tests with constant charge-discharge depth and charge and discharge rates throughout all charge-discharge cycles. In the actual world, however, charge-discharge depth and charge-discharge rates for each cycle are scarcely consistent and vary substantially over time for the same user as well as for various users who use their devices in different ways. The key challenge for such advanced battery models is estimating the SOH of batteries in real-world scenarios with such random fluctuations in charge-discharge depth and C-rates. For that purpose, the current study performs accelerated life cycling experiments on batteries with random variations in these two parameters, both separately and together. During the randomized charging process, several upper and lower bounds for state of charge variation were used to imitate users who use their smartphones in varied ranges of SOC, i.e. 0 – 40%, 60 – 100%, 30 – 70%, and 0 – 100%, and C-rates ranging from 0.5C to 2.5C. The SOH estimation models used in previous studies, which used fixed depth of charge/discharge and C-rate battery deterioration data, were used to estimate the SOH of the batteries tested in a randomized method to test their robustness. Finally, multiple iterations of the SOH estimation models have been presented with different predictor variables so as to minimize the model validation error.

Presenting Author: Ved Soni Auburn University