How Machine Learning can Improve Offshore Wind Farm Performance

Published 21 March 2019

A Catapult engineer investigates the structural integrity of a turbine blade using Wideblue software as part of the BOHEM project.

By Peter van Heck, Data Scientist in the Catapult’s Data & Digitalisation team.

The application of machine learning is revolutionising the sector’s approach to operations and maintenance.

A form of artificial intelligence, machine learning involves using computer systems to analyse vast quantities of data. The machine recognises patterns and trends from that data, from which we are then able to extract valuable information and learning.

By extracting value from data in this way, operators can gain a competitive advantage. For global technology firms, data is the raw material that drives their business. Offshore renewables organisations must follow their lead or risk being left behind in the Fourth Industrial Revolution.

Wind farms generate vast quantities of data in a variety of forms. Onboard SCADA systems produce operational data covering every aspect of the turbine. In addition, a wind farm’s support infrastructure generates data in the form of work orders, supply chain management data and the like. Currently, the Catapult observes that much of this data is simply archived – or worse, discarded.

This data has the potential to offer valuable insights, and machine learning is a highly effective technique to uncover them. Many of the current technology challenges in offshore renewables are solved by manual, laborious inspection, or even by guesswork. For example, identifying leading edge erosion of turbine blades is currently done by shutting the turbine down and inspecting the blade. The Catapult is exploring the use of machine learning to identify erosion using SCADA data and drone footage.

Machine learning can also be used for predictive maintenance. By identifying when a component deviates from normal operation, operators may be able to take corrective action before major damage occurs. The application of machine learning can be applied to many other forms of data: text, image, video and sound data are all ripe for analysis.

By combining the industry expertise of engineers, managers and business analysts with data science techniques, offshore renewables organisations will be uniquely placed to benefit from the digital transformation. Operators need to review their data management procedures, with a focus on making data available to analysts, and then must then work to understand the data that their organisations have and create infrastructure that supports machine learning.

The Catapult’s experience is that short pilot projects, combining field experts of both offshore renewables and data science, are a good way to start and show the potential of machine learning. For more information, visit our Data & Digitalisation Services page and contact our team to start your journey towards a more effective, data-driven future.