Timezone: »

Learning to Identify Drilling Defects in TurbineBlades with Single Stage Detectors
Andrea Panizza · Szymon Tomasz Stefanek · Stefano Melacci · Giacomo Veneri · Marco Gori

Nondestructive testing (NDT) is widely applied to defect identification of turbine components during manufacturing and operation. Operational efficiency is key for gas turbine OEM (Original Equipment Manufacturers). Automating the inspection process as much as possible, while minimizing the uncertainties involved, is thus crucial. We propose a model based on RetinaNet to identify drilling defects in X-ray images of turbine blades. The application is challenging due to the large image resolutions in which defects are very small and hardly captured by the commonly used anchor sizes, and also due to the small size of the available dataset. As a matter of fact, all these issues are pretty common in the application of Deep Learning-based object detection models to industrial defect data. We overcome such issues using open source models, splitting the input images into tiles and scaling them up, applying heavy data augmentation, and optimizing the anchor size and aspect ratios with a differential evolution solver. We validate the model with 3-fold cross-validation, showing a very high accuracy in identifying images with defects. We also define a set of best practices which can help other practitioners overcome similar challenges.

Author Information

Andrea Panizza (Baker Hughes)
Szymon Tomasz Stefanek (Siena Artificial Intelligence Laboratory)
Stefano Melacci (University of Siena)
Giacomo Veneri (Baker Hughes)

Giacomo Veneri (Siena on 11/17/1973) graduated in Telecommunications Engineering in 1999. From 1999 to 2004 he collaborated with the University of Siena as an expert of Human Computer Interaction in the context of Multimedia and Bioengineering. Since 2004 to 2006 he collaborated with major international financial institutions in the field of trading and in information visualization. Since 2006 to 2012 Technical Director and Director of the research laboratory (accredited MIUR) of [Etruria Innovation SPA]. PhD graduated at the Department of Neurological Sciences, University of Siena. It deals with human computer interaction and human mechanisms that underlie human-machine interaction. He is Java 7 OCA Associated certified, Predix certified, GreenBelt certified, SCRUM certified. Author of Maven Build Customization and Hands-On Industrial Internet of Things.

Marco Gori (University of Siena)

More from the Same Authors