LSTM-RNN based PV sources integration with CC–CV charging for sustainable electric mobility
- 1Department of Electrical and Electronics Engineering, Chennai Institute of Technology
- 2Department of Electrical and Electronics Engineering, College of Engineering, Guindy, Anna University
Abstract
The growing adoption of Photo Voltaic (PV) powered Electric Vehicle (EV) charging systems are hindered by solar intermittency and the absence of coordinated forecasting-driven charging strategies. Existing methods often decouple PV generation from EV demand or rely on static scheduling, leading to suboptimal energy utilization and increased grid reliance. Additionally, thermal effects influencing PV output remain underexplored. This study proposes a data-driven framework that integrates Long Short Term Memory–Recurrent Neural Network (LSTM–RNN) based surplus solar forecasting with adaptive EV charging coordination for a 300 kW rooftop PV system in Kundrathur, Chennai, India. The model predicts surplus PV energy at 30-minute intervals using time-series inputs such as irradiance, temperature, load demand, and state of charge. These forecasts guide dynamic scheduling of slow and Direct Current (DC) fast charging under a Constant Current–Constant Voltage (CC–CV) protocol. Validated using real operational data, the model achieves a Mean Absolute Percentage Error (MAPE) of 2.421%, Root Mean Square Error (RMSE) of 24.65 kWh, and Mean Absolute Error (MAE) of 16.85 kWh, demonstrating short-term forecasting accuracy. The system delivers a daily surplus energy of 588.97 kWh, supporting approximately 18 EVs (32 kWh battery) or 14 EVs (40 kWh battery), eliminating grid dependency during peak solar periods. DC fast charging reduces charging time from 5 h to 2 h, while controlled constant voltage tapering and thermal management ensure safe battery operation. Furthermore, each EV charged solely using solar energy avoids approximately 28 kg of CO₂ emissions, aligning with Sustainable Development Goals (SDGs), 7 (Affordable and Clean Energy), 11 (Sustainable Cities and Communities), and 13 (Climate Action). Overall, the framework demonstrates that integrating solar forecasting with thermally aware adaptive charging enhances renewable energy utilization in EV charging systems, while highlighting the need for future improvements under variable conditions.
Related
- E2C2AKA: enhanced ECC-based authentication and key agreement for securing communication in electric vehicle charging systems
- Food-Grade By-Products as Sustainable Sources of Phytochemicals: Green Extraction Strategies, Encapsulation Technologies, and Applications in Functional Food Development
- Research on resource sharing and allocation incentive mechanism of new energy vehicle charging cloud platform
- RSS-enabled hybrid-modulation 19-level switched-capacitor inverter with ninefold voltage boost inductorless soft-charging low device count and minimized LDP
- Measuring Mobility Function in Everyday Life—Beyond the Stopwatch
BibTeX
@article{murugesan2026lstmrnn,
title = {{LSTM-RNN based PV sources integration with CC–CV charging for sustainable electric mobility}},
author = {Ezhilmaran Murugesan and Somasundaram Periasamy},
journal = {Scientific Reports},
year = {2026},
month = aug,
doi = {10.1038/s41598-026-65919-1},
url = {https://doi.org/10.1038/s41598-026-65919-1},
note = {CC BY},
}
RIS
TY - JOUR
TI - LSTM-RNN based PV sources integration with CC–CV charging for sustainable electric mobility
AU - Ezhilmaran Murugesan
AU - Somasundaram Periasamy
PY - 2026
DA - 2026/08/01
JO - Scientific Reports
DO - 10.1038/s41598-026-65919-1
UR - https://doi.org/10.1038/s41598-026-65919-1
AB - The growing adoption of Photo Voltaic (PV) powered Electric Vehicle (EV) charging systems are hindered by solar intermittency and the absence of coordinated forecasting-driven charging strategies. Existing methods often decouple PV generation from EV demand or rely on static scheduling, leading to suboptimal energy utilization and increased grid reliance. Additionally, thermal effects influencing PV output remain underexplored. This study proposes a data-driven framework that integrates Long Short Term Memory–Recurrent Neural Network (LSTM–RNN) based surplus solar forecasting with adaptive EV charging coordination for a 300 kW rooftop PV system in Kundrathur, Chennai, India. The model predicts surplus PV energy at 30-minute intervals using time-series inputs such as irradiance, temperature, load demand, and state of charge. These forecasts guide dynamic scheduling of slow and Direct Current (DC) fast charging under a Constant Current–Constant Voltage (CC–CV) protocol. Validated using real operational data, the model achieves a Mean Absolute Percentage Error (MAPE) of 2.421%, Root Mean Square Error (RMSE) of 24.65 kWh, and Mean Absolute Error (MAE) of 16.85 kWh, demonstrating short-term forecasting accuracy. The system delivers a daily surplus energy of 588.97 kWh, supporting approximately 18 EVs (32 kWh battery) or 14 EVs (40 kWh battery), eliminating grid dependency during peak solar periods. DC fast charging reduces charging time from 5 h to 2 h, while controlled constant voltage tapering and thermal management ensure safe battery operation. Furthermore, each EV charged solely using solar energy avoids approximately 28 kg of CO₂ emissions, aligning with Sustainable Development Goals (SDGs), 7 (Affordable and Clean Energy), 11 (Sustainable Cities and Communities), and 13 (Climate Action). Overall, the framework demonstrates that integrating solar forecasting with thermally aware adaptive charging enhances renewable energy utilization in EV charging systems, while highlighting the need for future improvements under variable conditions.
SN - 2045-2322
ER -