As climate action and energy security agendas drive greater demand for renewable energy, wind power capacity continues to grow, as does the size of wind turbines. Because they are subject to external factors that cause mechanical failure, wind turbines require proactive monitoring and preventive maintenance to avoid costly repairs and downtime.

Artificial intelligence (AI) and machine learning (ML) are changing the game when it comes to renewable asset management.

At WindEurope 2023 in Copenhagen, CGI renewable energy expert Rita Burnay presented a poster on a CGI-supported study of an ML-powered system for predicting wind turbine failures, connected with CGI’s Renewables Management System (RMS) for asset monitoring. 

As documented in this study, the powerful prediction capabilities of AI have a direct impact on asset management efficiency by helping to reduce fault events and failure time. This reduces downtime, improves the longevity of components, lowers maintenance costs and increases long-term profit from energy production.

Developing reliable models for predicting performance

The study uses historical data from multiple signals to build binary classification models for predicting wind turbine faults. Tracking 11 fault scopes over a three-month forecast, one of the main goals was to ensure reliability of the models, with as few false positives as possible.

Using Microsoft Azure Machine Learning Studio, models were developed for each fault that recurred among the wind turbines. Each trained and fitted model received data from the wind turbines and classified the time window as a possible failure prediction in that particular wind turbine.

The result? The models behaved reliably and did not present any false positive predictions. Further, they were able to predict at least one failure event for six fault scopes.

While this study focused on wind energy, its foundation and AI methodology are applicable for other renewable energy sources, such as solar.

Explore the power of AI for renewable asset management with CGI

Organizations need data-driven insights to assess and improve operational performance, and to enable strategic analysis and decision-making. We help clients move from supervision to hyper-vision, advancing beyond monitoring assets toward making intelligent predictions.

With CGI’s RMS, a monitoring system that analyzes and operates the assets, clients are able to combine their portfolio into a single solution and optimize efficiency. Connecting the intelligent failure prediction tool to CGI RMS allows them to analyze the top faults of each asset and activate the prediction mechanism.

As AI-powered predictive tools continue to drive operational excellence across renewables, we are excited to be at the forefront of realizing the possibilities they present.