Robots and AI Beat Humans by 14 Pct in Detecting Turbine Fault in Innovate UK Project

Perceptual Robotics and the University of Bristol have announced the results of an Innovate UK data analysis project which show that robots and artificial intelligence (AI) are 14 per cent more accurate in detecting faults in wind turbines than expert humans carrying out the same inspections.

Perceptual Robotics
Source: Perceptual Robotics

The Innovate UK Research and Development project involved incorporating fully-automated surface defect detection into the data-processing pipeline for wind turbine inspections.

According to the partners, while the capture of images taken during inspections has previously been automated, it is the first time the processing of the images was carried out fully automated.

“Until now, wind turbine operators have been left in the dark about the capabilities of fully-automated inspections when compared with manual ones as there has been a lack of benchmarking to show comparisons between the two,” said Kostas Karachalios, CEO of Perceptual Robotics.

“We have shown incorporating fully-automated surface defect detection into our Dhalion system enhances the reproducibility and speed of current wind turbine inspections, significantly reducing costs, increasing quality and reducing safety concerns.”

The ongoing project between Perceptual Robotics and the University of Bristol was initially for two years but has been extended by one year with DNV.

Related Article

The first two years of the project focused on offshore wind turbine inspections before being extended for a further year to consider validation of results in both onshore and extreme, offshore environments.

The project focused on demonstrating the capabilities of the inspection system by performing end-to-end validating and verifying of the data system.

The project partners also worked on determining how the data collection process can be auditable and traceable, and on analysing the way the performance is measured to ensure it was as accurate as possible.

The involvement of DNV in the project’s extended third year allowed the partners to objectively assess the inspection system and data, according to Perpetual Robotics.

Related Article

The UK-based company said it was also able to reduce its data review turnaround time by 27 per cent which allows receiving inspection results within 48 hours.

Follow on: