Probabilistic operational management of a renewable-based microgrid considering uncertainties using the self-adaptive gravitational search algorithm

Keeping the Lights On When the Future Is Uncertain

As more homes and businesses draw power from rooftop solar panels, local wind turbines, and batteries, keeping the lights on at a fair price becomes a tricky balancing act. Sunshine, wind, and electricity prices all change from hour to hour, and small neighborhood-sized power systems called microgrids must react in real time. This paper explores a new way to run such microgrids so they stay reliable, make the most of clean energy, and keep costs down even when the future is uncertain.

Small Power Networks Close to Home

A microgrid is essentially a mini power network serving a campus, neighborhood, or small town. Instead of relying only on a distant power plant, it can draw from nearby sources such as a microturbine, fuel cell, solar panels, wind turbine, and a battery bank, while still connecting to the main grid when needed. A central controller decides, hour by hour, how much power each device should produce or store. The challenge is that this controller must make choices based on predictions of how much people will use, how strongly the wind will blow, how brightly the sun will shine, and what the market price of electricity will be. If these guesses are wrong, the microgrid may end up paying too much or risking shortages.

Making Smart Guesses About an Uncertain Future

Many existing control methods either ignore uncertainty or try to handle it with heavy-duty statistical tools that are too slow for real-world operation. The authors turn to a lighter-weight technique called the 2m-Point Estimation Method. Instead of running thousands of random simulations, this method picks just a small number of carefully chosen “what if” points around the expected values of key inputs, such as demand, renewable output, and market price. By solving the microgrid planning problem at these points, it can estimate how total cost and power flows will behave under a wide range of conditions. This gives the controller a practical picture of risk and variability without overwhelming computing resources.

A Gravity-Inspired Search for Better Schedules

To decide the best schedule for all generators, the battery, and the grid connection, the study uses a search method inspired by gravity. In this scheme, each possible schedule is like a particle with a mass related to how good that schedule is: lower cost and better use of renewables mean a heavier mass. These particles “pull” on each other, drifting toward better regions of the solution space. The authors improve this idea with a self-adaptive twist. They introduce two types of random nudges, or mutation moves, that help the particles escape bad local traps and explore more widely. The algorithm tracks which move works better over time and automatically shifts its behavior, giving more weight to the more successful strategy. This self-tuning approach, called the Self-Adaptive Gravitational Search Algorithm, helps it find high-quality schedules faster and more reliably.

Putting Batteries and Smarter Control to the Test

The team tests their combined framework—using point estimation for uncertainty and the improved gravity-based search—on a model low-voltage microgrid with solar, wind, a microturbine, a fuel cell, and a nickel–metal hydride battery. They compare two main cases: one without a battery and one with a battery under their new control method. When the battery is included, the controller can shift energy in time, charging when power is cheap or plentiful and discharging when demand or prices spike. In simulations, this cuts the total daily generation cost almost in half and increases the share of renewable energy by about 10 percentage points during peak hours. The upgraded algorithm also converges about a quarter faster and gives lower operating costs than standard gravity-based methods and a widely used particle swarm approach.

Robust Performance in a Changing World

To probe how sturdy the method is, the authors vary both demand and market prices up and down by 10 and 20 percent. Even in the worst case, when both are 20 percent higher than expected, total costs rise only modestly, and the drop in renewable use is limited. Competing methods show larger cost swings and are more easily thrown off. The study also explores future scenarios where falling prices for solar, wind, and advanced batteries would further boost savings and clean energy use. Overall, the work suggests that combining smart uncertainty modeling with a self-tuning search algorithm can help microgrids deliver cheaper, cleaner, and more reliable power, even when tomorrow’s weather and prices are anyone’s guess.

Citation: Ahmed, E.M., Zaki, Z.A., Kamarposhti, M.A. et al. Probabilistic operational management of a renewable-based microgrid considering uncertainties using the self-adaptive gravitational search algorithm. Sci Rep 16, 12313 (2026). https://doi.org/10.1038/s41598-026-42839-8

Keywords: microgrid, renewable energy, battery storage, optimization algorithm, energy management