Investigation of Loading Variation Effect on Lithium Iron Phosphate Battery Electrodes Using Long Short Term Memory

Due to their high energy density, high discharge capability, and extended cycle life, lithium iron phosphate batteries (LiFePO4) are excellent for use in heavy-duty applications such as energy storage systems (ESS), electric vehicles (EV), and micro-grids. However, stress development in electrode materials because of unfavorable charge and discharge conditions leads the battery cell to degrade quickly. As a result, precise prediction of the state of charge (SOC), state of health (SOH), and capacity degradation of the battery has become critical. The goal of this project is to investigate the load fluctuation effect during various charge rates (DCR) and different discharge rates (DDR) utilizing the battery's electrode. A deep learning algorithm (DLA) based model is developed in this research by tuning hyperparameters to forecast battery performance under varied loading instances. A recurrent neural network with long-short term memory (RNN-LSTM) with 10 neurons in the hidden layer predicted battery performance with the lowest percentage of error based on mean squared error (MSE) and root-mean squared error (RMSE). Low error levels $(\lt0.1$%) under diverse load conditions suggested that the model is robust and capable of modelling with high accuracy despite loading variation in the data. Finally, the analysis of the loading effect on energy storage found that the Li-FePo4 battery is the best solution compared to other batteries because it shows slow ageing with minimal degradation even when loaded above the maximum Crate.