Model Predictive Energy Management for PV- Battery Systems in Smart Microgrids


Creative Commons License

Cıtlı A. H., Yucel O. Y., Fakıoglu F., Onal R., Kayışlı K., Düzkaya H., ...Daha Fazla

2026 International Conference on Artificial Intelligence Research and Application (ICAIRA), İstanbul, Türkiye, 2 - 04 Eylül 2026, ss.1421-1426, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/icaira72375.2026.11769099
  • Basıldığı Şehir: İstanbul
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.1421-1426
  • Gazi Üniversitesi Adresli: Evet

Özet

The integration of renewable energy sources and volatile electricity demand poses significant operational challenges for institutional microgrids. Conventional rule-based energy management systems fail to adapt to dynamic market pricing and often ignore battery State of Health (SoH) dynamics, leading to high utility costs during peak hours and premature hardware degradation. This paper presents a constraint-aware Model Predictive Control (MPC) framework specifically tailored for the Gazi University Faculty of Engineering. The proposed system centrally optimizes a 900 kWp rooftop photovoltaic (PV) array and a 400 kWh Lithium Iron Phosphate (LFP) battery energy storage system (BESS). By ingesting dynamic EPİAŞ Day- Ahead Market prices and generating short-term load forecasts, the MPC algorithm actively schedules optimal charge and discharge cycles over a 15-minute sampling interval. Crucially, the mathematical solver incorporates strict SoH degradation penalties and explicit 20%-90% State of Charge (SoC) constraints to preserve battery longevity. Based on 2025 faculty operational data (annual load: 1,608.09 MWh; PV generation: 1,292.42 MWh/year), the MPC strategy successfully mitigates peak demand fees during high-tariff winter periods and manages a seasonal summer surplus of approximately 212.23 MWh/year. Furthermore, the environmental impact assessment confirms an annual carbon-emission avoidance potential of 674.3-806.7 tCO₂/year. Ultimately, the framework establishes an economically viable and scalable smart grid solution that significantly reduces utility overhead costs while actively extending the operational lifespan of the battery assets. The new thing about this framework is that it puts penalties for getting old into the part of the system that makes real-time decisions. This means it provides a precise, site-specific solution that is validated against real campus data. The framework focuses on finding a solution that works well for the campus, so it uses real campus data to ensure its effectiveness. The campus data is what makes the framework's solution so accurate.