An improved material-inspired generation algorithm for load frequency control in EV-integrated power systems

Why keeping the lights steady matters

Every device in our homes, from refrigerators to phone chargers, depends on the electric grid holding a nearly constant frequency. As more electric vehicles (EVs) plug in and more regions share power over long distances, keeping that frequency steady becomes harder. This study explores a new way to fine‑tune grid controllers so they can juggle traditional power plants and huge fleets of EVs while keeping the system stable and reliable.

Electric cars as helpers, not just users

Modern power grids are increasingly interconnected, with large areas sharing electricity through tie‑lines. At the same time, millions of EVs are appearing as flexible, fast‑responding loads. EVs do not just consume power; when connected through suitable electronics, they can briefly push power back into the grid, helping to steady frequency after a disturbance such as a sudden jump in demand. However, EV charging and driving patterns are unpredictable, and renewable sources like wind and solar add further swings. Traditional control schemes that rely on simple tuning rules often struggle with this mix of uncertainties, leading to unwanted oscillations in grid frequency and power flows between areas.

A chemistry‑inspired way to search for better settings

The authors build on a recent "material generation" algorithm that borrows ideas from how atoms form stable chemical compounds. In this view, each possible set of controller settings is like a material made from different elements. The algorithm forms, breaks, and recombines these virtual materials, judging their "stability" by how well they keep grid frequency and tie‑line power within limits in computer simulations. But the original version of this method had a weakness familiar from many search techniques: it could settle too early on a merely adequate solution and stop exploring better ones, especially in complex, high‑dimensional problems.

Adding a smarter curve to avoid getting stuck

To overcome this, the study introduces an Improved Material Generation Algorithm (IMGA). Alongside the chemistry‑inspired moves, IMGA periodically performs a small, local "curve fit" step: it looks at three neighboring solutions and fits a simple curved line (a parabola) through their performance. From this curve it estimates a better nearby point to test next. This quadratic interpolation step gives the search a sense of direction, letting it zoom in on promising regions without relying on a single best solution as a leader. A probabilistic switch decides when to use this curve‑based move versus the original random recombination, preserving diversity in the pool of candidates and helping the search escape local dead‑ends.

Testing the idea on a shared grid with EV fleets

To see whether this smarter search pays off in practice, the authors model a grid with two connected areas, each supplied by conventional thermal power plants and large fleets of EVs. They use a layered controller: an outer loop responds quickly to sudden changes, while an inner loop removes any steady‑state error. IMGA tunes the gains of this cascaded controller so that, in simulation, frequency deviations and tie‑line power swings are as small and short‑lived as possible. The improved algorithm is compared with several other well‑known search methods, including particle swarm optimization and other modern metaheuristics, under a range of scenarios: with and without EV participation, disturbances in one area or the other, and irregular, step‑like changes in demand.

What the simulations reveal

Across all cases, IMGA consistently finds controller settings that make the grid respond more smoothly than the alternatives. It converges faster and more reliably, showing smaller spread between best and worst runs. With EVs actively helping, the tuned controller reduces the depth of the initial frequency dip by up to about half and shortens the time the system needs to settle by a few seconds, compared with operation without EV support. Under arbitrary load changes, the IMGA‑tuned controller keeps both frequency and power exchange between the two areas closer to their desired values than a previously proposed design, damping oscillations more quickly and limiting overshoot.

What this means for future grids

In plain terms, the study shows that a smarter, chemistry‑and‑curve‑inspired search method can find better ways to adjust grid controllers so that they make effective use of EV fleets as stabilizing partners. The improved algorithm delivers tighter control of frequency and smoother power sharing between regions in simulations, even when conditions are noisy and changing. While the EV and grid models are still simplified and tested on a two‑area system, the results suggest that such advanced tuning tools could help future, EV‑rich power systems stay steady and reliable without costly over‑engineering.

Citation: Almutairi, S.Z., Ginidi, A.R. An improved material-inspired generation algorithm for load frequency control in EV-integrated power systems. Sci Rep 16, 13020 (2026). https://doi.org/10.1038/s41598-026-47360-6

Keywords: electric vehicles, power grid stability, frequency control, optimization algorithms, smart grids