Skip to content
Conference Open access

A Review on Advanced Data-Driven Models for Accurate and Dynamic Early Prediction of Electric Vehicle Battery Remaining Useful Life

Oct 2026 · International Conference on Innovative Mechanisms for Industry Applications · pp. 226-232 · 0 citations · 16 references

Abstract

The adoption of electric vehicles has increased rapidly. This growth has created a need for accurate battery health assessment because incorrect Remaining Useful Life (RUL) estimation can affect system reliability and safety. Traditional machine learning and deep learning methods have limitations in modeling complex temporal dependencies. They also require manual feature engineering, resulting in reduced performance under varying operating conditions. The researchers obtained battery data from the National Aeronautics and Space Administration (NASA) battery dataset, which includes essential data points such as voltage, current, temperature, and capacity, because they executed preprocessing tasks for noise elimination and outlier identification and Min-Max scaling. The selected features are transformed into time-series sequences and fed into the hybrid model. The GRU component captures short-term variations, while the LSTM component learns long-term degradation patterns. The hybrid GRU–LSTM architecture integrates both short-term and long-term temporal features to enhance prediction capability. The researchers used Glowworm Swarm Optimization (GSO) to optimize the model’s hyperparameters which resulted in better prediction accuracy and increased system performance. The model achieved prediction results which showed measurement accuracy through Mean Absolute Error (MAE) 0.2011 and Root Mean Squared Error (RMSE) 0.2250 and coefficient of determination (R2) 0.9996. The framework proved effective at reducing prediction mistakes while maintaining strong operational capabilities which makes it appropriate for EV practical applications battery management systems.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.