A Predictive Model for the Charging Capacity of Grid-Scale Lithium-Ion Battery Connected to Renewable Energy Sources
Open Access DepositedThis praxis investigates the charging capacity and predictive modelling of grid-scale lithium-ion batteries (LIBs) connected to renewable energy sources (RES). The study addresses key research questions concerning the influence of operational and environmental parameters on battery performance, the development of predictive models for battery capacity, and the effectiveness of Long Short-Term Memory (LSTM) models in enhancing prediction accuracy.The research specifically explores the effects of intermittent and variable charging currents on the charging capacity of LIBs which revealed a significant impact on battery performance. Environmental and cell temperature fluctuations impact on LIBs is also studied as well as their combined influence after showing a notable correlation between them. To address these findings, the study develops predictive models using advanced machine learning techniques. Initial models employing decision tree regressors with ensemble learning and eXtreme Gradient Boosting (XGBoost) exhibit limitations in capturing the downward trend of actual data despite their relatively low errors. Consequently, LSTM models are introduced due to their effectiveness in handling time-series data and long-range temporal dependencies. Various LSTM architectures are tested and the best performing model is identified. This model demonstrates strong generalization capabilities and accurately captures the degradation trend of LIB capacity over time. The study also explores the issue of multicollinearity by combining cell and environmental temperatures into an average temperature feature. However, this approach leads to a deterioration in model performance, highlighting the complexity of temperature interactions and their impact on predictive accuracy. The findings of this research offer valuable insights into the operational and environmental factors affecting the performance of LIBs in the field of renewable energy storage, the development of robust predictive models, and the practical application of LSTM networks for battery capacity prediction. Finally, the study underscores the importance of real-world data in enhancing model generalization ability and suggests future research directions.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
BouShakra_gwu_0075A_16987.pdf | 2025-04-11 | Open Access |
|