Skip to main content
MB Abdelghany

Optimal and data-driven control

Back to research overview

My control research asks how complex energy systems can make reliable decisions at the speed of the physical process without losing the longer view of markets, degradation, uncertainty, and infrastructure planning.

Research scope

I study optimal, robust, stochastic, economic, and data-informed control for cyber-physical energy systems. The work spans receding-horizon and multi-timescale model predictive control, hybrid-system validation, grid-forming inverter control, renewable microgrids, electrolysers, and hybrid storage.

Research objectives

The objective is to coordinate planning and real-time action while keeping physical constraints, stability, component health, and computational effort visible. Data is introduced where it improves prediction, state estimation, or adaptation; it does not replace the structure supplied by physics and control theory.

Control architecture

The multi-layer predictive-control architecture separates long-horizon scheduling from short-horizon correction. Forecasts, operating states, device constraints, and physical transients remain coordinated without being forced into one timescale.

Multilayer model predictive control diagram connecting energy forecasts and a hybrid energy-storage model to high- and low-layer controllers.
The high layer schedules the hybrid system from forecasts and operating states; the low layer tracks and corrects that decision in real time.Source: Abdelghany et al., IEEE Transactions on Industrial Informatics (2024), Fig. 8 (opens in a new tab) · CC BY-NC-ND 4.0

Experimental validation

Real-time simulation and converter hardware expose delay, saturation, measurement noise, current limits, and computation time, providing direct evidence of whether the controller remains credible beyond its numerical model.

ESCO laboratory grid-forming inverter testbed with real-time simulation, power converters, measurement equipment, and oscilloscopes.
The ESCO laboratory closes the loop between analysis and hardware through real-time simulation, converter control, measurement, and repeatable disturbances.

Measured performance

Renewable-generation and load profiles define the experimental operating problem, and the measured emulator response shows how closely the controller follows the requested demand across the complete profile, including its steepest changes.

Renewable generation profiles and load demand used in the experimental microgrid emulator.
Forecasted renewable profiles and the requested load define the operating problem presented to the experimental emulator.Source: Abdelghany et al., IEEE Transactions on Industrial Informatics (2024), Fig. 22 (opens in a new tab) · CC BY-NC-ND 4.0
Experimental load reference and measured response tracking closely over a 24-hour profile.
The measured response follows the requested load across the 24-hour experimental profile, including the steepest changes in demand.Source: Abdelghany et al., IEEE Transactions on Industrial Informatics (2024), Fig. 23 (opens in a new tab) · CC BY-NC-ND 4.0

Selected publications

Browse the complete publication record

Projects and facilities

  • Experimental environment

    ESCO Lab

    Real-time simulation, power converters, vehicle emulators, and measurement hardware provide the experimental layer for controller validation.

    View

  • Current funded project

    Autonomous renewable-hydrogen microgrids for Arctic communities

    Digital twins, model-predictive and fault-tolerant control, diagnostics, and resilience assessment are brought together for isolated polar infrastructure.

    View

Research opportunities

Expressions of interest are welcome from researchers working in predictive control, hybrid systems, learning-assisted control, grid-forming systems, and experimental validation. Availability and funding are confirmed individually.