Electric vehicles (EVs) operate on electricity stored in rechargeable lithium-ion batteries (LIBs). The state of charge (SOC) is a key aspect in determining how much energy is available in batteries. Accurate SOC estimation is crucial for predicting battery lifetime and ensuring safe EV operation. We describe a novel approach for accurate SOC estimation based on neural networks (NNs) built with the neural designer (ND) software. The training data was obtained from NASA Prognostics Centre of Excellence (PCoE) dataset. To determine the SOC, we demonstrated three NN models trained with three distinct optimizers. To estimate the SOC battery data such as voltage, current, and temperature were directly mapped to the NN models. For model accuracy, the suggested NN models have been deployed for all three battery datasets, namely B0005, B0006, & B0007, and these trained NN models are compared using the performance metrics root mean squared error (RMSE) and Minkowski error (MKIE). Finally, the NN models were extensively tested using the goodness-of-fit (GOF) technique.
A Novel Neural Designer-Based Machine Learning Models for State-of-Charge Estimation of Lithium-Ion EV Batteries
2024-07-31
879987 byte
Conference paper
Electronic Resource
English
Robust State of Charge Estimation of Lithium-Ion Batteries via an Iterative Learning Observer
SAE Technical Papers | 2012
|Robust State of Charge Estimation of Lithium-Ion Batteries via an Iterative Learning Observer
British Library Conference Proceedings | 2012
|Robust state of charge estimation of lithium-ion batteries via an iterative learning observer
Automotive engineering | 2012
|STATE OF CHARGE ESTIMATION BASED ON STATE OF HEALTH CORRECTION FOR LITHIUM-ION BATTERIES
British Library Conference Proceedings | 2018
|