Predict4Resilience:  Leveraging AI and Weather Data to Strengthen Grid Resilience 

Weather plays a key role in how the electricity network operates. Extreme weather events are responsible for most of the disruptions in the energy supply with evidence showing that climate change has contributed to more frequent extreme weather events. Currently estimating the impact of severe or extreme weather relies on user judgment and is subject to human bias and experience. 

P4R, trialled in partnership with SP Energy Networks (SPEN), is a fault forecasting system using multiple complex data sources with novel statistical learning, that will provide short-term fault forecasts and early warning of fault volumes up to 5 days ahead.

It uses probabilistic fault prediction and related decision-support for the first time in a GB innovation project, transforming human-centric decision-making and improving the response to faults on the HV network.

By helping Control Room teams take proactive, data-driven action ahead of severe weather, they can ensure equipment and engineers are in place to tackle faults quicker than ever before. Importantly, P4R does not automate decisions directly on the network, but rather augments human decision-making, ensuring operators remain in control while benefiting from enhanced situational awareness.

This proactive response means power supply can be restored faster than is currently possible, reducing the time customers are without power and generating direct benefits to consumers, network operators and the environment while creating a more resilient network for now and the future as we move towards Net Zero.

This represents a major shift from human interpretation of basic weather forecasts to sophisticated data-driven analytics. Results have shown that P4R forecasts are actionable and expected to lead to multiple immediate benefits, which will be enhanced by continuous improvement.

The project was delivered in a short timeframe (build and deployment within a year), leveraging a combination of pre-existing accelerators, domain expertise, and custom model development. It also required significant change management, as it transformed how operational teams plan interventions — particularly in terms of pre-positioning field crews ahead of anticipated events, improving response times while reducing unnecessary mobilization costs.

Ultimately, Predict4Resilience illustrates how AI, data, and user-centric design can be combined to enhance the resilience of critical infrastructure. As climate-related events become more frequent and severe, such capabilities are expected to play an increasingly central role in the evolution of energy network operations.