An international research team has developed an artificial intelligence approach to help plan renewable energy systems that remain reliable under real-world climate variability, including extreme weather events. The method, described in the journal Progress in Energy, aims to address a key weakness in traditional energy planning: models that often fail to capture rare but disruptive conditions, leading to systems that are either too costly or not dependable enough.
The researchers built an AI model trained on three decades of wind and solar data from Pingtan, an island region in China known for its variable weather. The model generated thousands of realistic weather scenarios, ranging from typical seasonal patterns to severe storms and prolonged cloudy or windless periods. These scenarios were then used to test different configurations of an Integrated Energy System, which included solar panels, wind turbines, batteries, hydrogen production and storage, and a connection to the national electricity grid.
By simulating how each system configuration would perform across all the AI-generated scenarios, the team identified the most cost-effective and resilient design. The optimal setup combines multiple technologies: wind and solar serve as the primary electricity sources, batteries manage short-term fluctuations in supply and demand, hydrogen storage provides backup during extended periods of low renewable generation, and the grid acts as an additional safety net.
A central finding is that batteries and hydrogen play complementary roles. Batteries are best suited for balancing daily variations in generation and consumption, while hydrogen is more effective for storing energy over weeks or months. Together, they enable a renewable system that is both reliable and economically viable, according to the researchers.
The study addresses a growing challenge for energy planners worldwide. As countries increase their reliance on solar and wind power, the inherent unpredictability of weather becomes a more significant factor in grid stability. Traditional planning methods often rely on historical averages, which can underestimate the frequency or severity of extreme events, leading to systems that may fail during heatwaves, cold snaps, or prolonged calm periods.
The AI-driven scenario framework offers a way to stress-test energy systems against a wide range of possible futures. By generating thousands of plausible weather sequences, the model allows planners to evaluate how different technology mixes would perform under conditions that have not necessarily occurred in the past but are physically possible. This is particularly important as climate change alters weather patterns and increases the likelihood of extreme events.
The researchers note that their approach can be adapted to other regions and scales, providing a tool for policymakers and utility companies to make more informed decisions about investments in renewable energy infrastructure. The study was led by Jiawei Tan and colleagues, with the full paper published in Progress in Energy.
