A Systematic Review of Hybrid Model-Based and Data-Driven Techniques for Real-Time State Estimation in Renewable Energy-Integrated Microgrids

Olaide Ayokunmi Oladosu¹*, Faithfulness Issijude², Allwell Chinedum Ihejirika³, Emmanuel Akukula Attarbo⁴, Elizabeth Emili⁵, Chukwunedu Ifechukwude⁵, Micheal Abimbola Oladosu⁶** and Moses Adondua Abah⁶

¹Department of Computer Science, Faculty of Science and Technology, Babcock University, Ilisan-Remo, Ogun State, Nigeria.

²Department of Computer and Robotics Education, Faculty of Vocational and Technical Education

University of Nigeria, Nsukka, Nsukka, Nigeria

³Department of Electrical Engineering, Faculty of Engineering Technology, Nigerian Defence Academy

Kaduna, Nigeria

⁴Electrical Engineering and computer science, South Dakota School of Mines and Technology, South Dakota, USA.

⁵Department of Computer Science, Faculty of Science, Anchor University, Lagos, Nigeria

⁶Department of Research and Linkages, Research Hub Nexus Institute, Ayobo, Lagos, Nigeria

(✉) Corresponding Author

Received: Apr 20, 2026/ Revised: June 2 2026/Accepted: June 4, 2026

Highlights

Systematically reviews hybrid model-based and data-driven state estimation techniques for microgrids.
Evaluates EKF–ML, physics-informed neural networks, and federated learning frameworks.
Highlights improved estimation accuracy and real-time performance in renewable-integrated microgrids.
Compares the strengths of physical modeling and machine learning-based approaches.
Identifies future challenges in cybersecurity, multi-energy integration, and benchmarking standards.

Abstract

The rapid proliferation of renewable energy sources (RES) within microgrids introduces substantial volatility and uncertainty that challenge conventional state estimation (SE) paradigms. Model-based approaches, while theoretically sound, often fail to capture the nonlinear dynamics introduced by photovoltaic arrays and wind turbines. Data-driven methods offer flexibility but lack the physical interpretability necessary for reliable real-time control. This systematic review critically appraises hybrid SE techniques that integrate model-based and data-driven approaches to achieve robust, real-time SE in renewable energy-integrated microgrids. Following PRISMA guidelines, 35 peer-reviewed studies published between 2020 and 2025 were identified and analysed from Scopus, Web of Science, and IEEE Xplore databases. Hybrid techniques, particularly physics-informed neural networks (PINNs), Extended Kalman Filter (EKF)-machine learning (ML) fusions, and federated learning frameworks, demonstrated superior accuracy, with root mean square errors (RMSE) below 0.015 per unit and real-time processing latencies under 50 ms. Hybrid SE frameworks represent the most promising trajectory for scalable, resilient microgrid state estimation, with critical gaps remaining in cybersecurity robustness, multi-energy system integration, and standardised benchmarking.

Keywords: State estimation; microgrids; hybrid methods; machine learning; Kalman filter; physics-informed neural networks; renewable energy; real-time systems

How to cite this article

Oladosu, O. A., Issijude, F., Ihejirika, A. C., Attarbo, E. A., Emili, E., Ifechukwude, C., Oladosu, M. A., & Abah, M. A. (2026). A systematic review of hybrid model-based and data-driven techniques for real-time state estimation in renewable energy-integrated microgrids. Science Archives, 7(2), 133–142. https://doi.org/10.47587/SA.2026.7212

This work is licensed under a Creative Commons Attribution 4.0 International License