The establishment of wind farms has been constantly growing during the past years as there is a need to harvest wind power so that people can be less dependent on fossil fuels. The number and the size of the wind turbines used to generate clean energy increases thus creating an urgent need to identify defective ones and conduct repair works.
Wind turbines may present several defects which are frequently associated with material damage during the manufacturing processes. The damage may not be initially detected but it can expand to a concerning degree when the blades of the turbines are subjected to complex dynamic loads during their operation.
Defects on wind turbines can impact their efficiency. Moreover, structural integrity issues may emerge if they are not adequately maintained. Nowadays, the inspection of the turbines is conducted via conventional techniques that include in-situ inspections and manual assessment of photographs. However, these methods are time-consuming and are also affected by other factors (e.g., lighting, subjectivity) and, therefore, do not always provide the optimum results. Other techniques such as ultrasound or visual thermography have also not proved to be adequate.
A team from the Loughborough University in the UK has recently introduced a new method that utilizes AI to detect the defects of the turbines via photographic analysis. The new tool utilizes the Mask R-CNN algorithm which is developed for image segmentation processes and it can detect erosion, cracks, voids or other defects in the turbines. Mask R-CNN is fed by images or videos captured either via in-situ inspection or via drone missions. The tool has been trained by a set of more than 900 images.
After the training, the AI tool was tested via a new set of images (223) where the defects of the turbines were already detected. The tool managed to recognize 85% of the defects proving its efficiency. Nevertheless, the team is aiming at further increasing the accuracy of the tool by improving the algorithm and by better pre-processing the images retrieved in-situ. Scientists admit that such AI-based tools are still underdeveloped and huge ameliorations will soon emerge. “AI is a powerful tool for defect detection and analysis, whether the defects are on wind turbine blades or other surfaces. Using AI, we can automate the process of identifying and assessing damages, making better use of experts’ time and efforts," Dr. Georgina Cosma, co-author of the study and a Senior Lecturer at the Loughborough University, stated.
Sources: LoughboroughUniversity Zhang et al. (2021)
