Numerical and experimental investigation of state of health of Li-ion battery

Estimation of State of Health (SoH) of Lithium-ion (Li-ion) battery is essential to predict the lifespan of batteries of an electric vehicle (EV). The efficient prediction of battery health indicates to the effective and safe operation of EV. However, delivering an effective and accurate method for the estimation of SoH in the real condition is truly a challenging task. The present study proposed a holistic procedure of combining both experimental and numerical investigations to conduct the fundamental study on coupled mechanical-electrochemical behavior of Li-ion battery. The proposed investigation highlighted the effect of stress on the capacity of the battery, considering capacity fade as an equivalent parameter to its health for real-time estimation of SoH. Finally, a simple model of Artificial Neural Network (ANN) is provided, which shows the linear dependency of stress with the SoH. The results obtained from the ANN model are validated with a Linear Regression (LR) model for a better understanding of the inspection. The predicted value of mean Square Error (MSE) and R square error in the ANN training model are found to be 0.000309 and 0.849687, respectively. Whereas for the test model, these predicted values are found to be 0.000438 and 0.819347, respectively.