An optimization framework for battery health management in vehicle-to-grid systems integrating transformer-based degradation prediction and grid service requirements

Why Smart Charging Matters

As electric cars become more common, they are not just vehicles anymore—they are rolling batteries. When plugged in, these batteries can send power back to the grid, helping to keep the lights on and smooth out the ups and downs of solar and wind power. This idea, called vehicle‑to‑grid, sounds like a win–win, but there is a catch: every extra charge and discharge slowly wears out the battery. This paper asks a practical question that many drivers, utilities, and city planners care about: can we use parked cars to support the grid and still protect battery life and owners’ wallets?

Turning Cars into Helpful Neighbors

In a vehicle‑to‑grid setup, parked electric cars act like a neighborhood energy reservoir. When electricity is cheap and plentiful, they charge up; when demand spikes and prices rise, they can feed energy back. The authors point out that most previous work tried to minimize electricity bills or satisfy grid needs, but treated battery wear as an afterthought. That is risky, because deep discharges, rapid power swings, and extreme states of charge can dramatically shorten a battery’s useful life. This study instead puts battery health at the center of the planning problem, while still aiming for strong economic returns and reliable grid support.

Teaching a Model to Feel Battery Wear

To protect the battery, the team first needs a way to predict how quickly it is aging under different usage patterns. Rather than relying on complex chemistry equations, they use a modern data‑driven approach inspired by tools from language translation. A “Transformer” model, better known for reading long sentences, is retrained to read long histories of battery behavior: the charge level over time, how quickly current flows in or out, the temperature, and how much capacity has already been lost. By learning from detailed test data from a university battery lab, the model can forecast how much extra wear will result from a proposed schedule of charging and discharging, even when effects build up slowly over many hours or days.

Balancing Health, Stability, and Profit

These battery wear predictions are then plugged into an optimization framework that searches for the best charging plan over a full day. The plan must juggle three goals. First, it should keep added battery damage as low as possible. Second, it should avoid sudden jumps in power, which can stress both the battery and the grid. Third, it should make money by buying electricity when prices are low and selling when they are high under time‑of‑use tariffs. The authors treat this as a nonlinear puzzle with many limits: power must stay within safe ranges, the charge level cannot go too high or too low, and changes from one hour to the next must not be too abrupt. To solve this puzzle efficiently and reliably, they use a mathematical method called sequential quadratic programming, which uses gradient information—how wear changes with small tweaks in power—to quickly home in on a feasible and near‑optimal schedule.

How the Smart Schedule Performs

The researchers test their method on a realistic electric vehicle battery and a 24‑hour price pattern. They compare three optimization tools: their chosen sequential quadratic programming method, another gradient‑based method, and a popular heuristic called particle swarm optimization, which explores options by imitating a flock of birds searching. All three respect battery safety limits, but the differences in performance are striking. The gradient‑based methods consistently find plans that charge aggressively when prices are low and discharge strongly when prices are high, using about 70% of the battery’s capacity range without over‑stress. This yields net profit while keeping estimated extra wear very small. The particle swarm approach, by contrast, often settles on timid charging and discharging, missing the most valuable moments and even losing money on average, especially when price patterns are varied at random.

What This Means for Drivers and the Grid

For a layperson, the take‑home message is reassuring: with smarter planning, parked electric vehicles can help stabilize the grid and earn income without dramatically shortening battery life. By combining an advanced prediction model that “understands” how batteries age with an optimizer that respects physical limits and price signals, the authors show that it is possible to find schedules that are both profitable and gentle on hardware. Their results suggest that utilities and charging operators should move beyond rule‑of‑thumb strategies and adopt health‑aware scheduling if they want vehicle‑to‑grid programs to be both sustainable and attractive to drivers in the long run.

Citation: Zhong, C., Ma, Q., Ren, M. et al. An optimization framework for battery health management in vehicle-to-grid systems integrating transformer-based degradation prediction and grid service requirements. Sci Rep 16, 12258 (2026). https://doi.org/10.1038/s41598-026-38862-4

Keywords: vehicle-to-grid, battery degradation, electric vehicles, smart charging, energy optimization