89 Enhancing EMAT Crack Detection Services Using State of the Art Deep Learning

Abstract

EMAT crack detection technology is used worldwide by many oil and gas operators to detect and size cracks in liquid and gas pipelines. By collecting data of increasingly higher resolution and quality, it is possible to achieve a more and more accurate representation of the integrity reality of oil and gas structures.

 

However, the prerequisite for gaining insights from this data is provided by Artificial Intelligence (AI) methods and a corresponding Research-First structure of a company. The AI division at ROSEN Research Center invested significant time and resources to apply powerful machine learning in particular Deep Learning methods to ensure timely and accurate inspection results delivered to operators. These efforts are directed towards building data driven applications to ensure reliability of inspection systems. Supported by a modern data engineering infrastructure, AI-powered data-driven applications can be used to enhance both the quality and efficiency of inspection data analysis.

 

This paper will provide an insight on how ROSEN uses AI-methods to enhance classification and sizing of metal-loss and crack indications in pipeline inspection data. Particularly, how Deep Learning methods help ensuring the quality of EMAT-Crack detection services.

KEYWORD(S) FOR SUBJECT AREA: Machine Learning, EMAT Technology

Read/Download PaperGradeAvailable after the event