Automatic Feature Prioritization for Ultrasonic Metal Loss Analysis Using AI

UltrasoundRefers to sound waves with frequencies above the upper limit of human hearing, typically above 20,000 Hz. Ultrasound is used to determine wall thickness and to … In-Line InspectionThe systematic process of visually examining, testing, or monitoring pipeline components to assess their condition, integrity, and compliance with regulatory re… is a widely used technology to gain high-quality information about possible pipeline threats like corrosionThe deterioration or degradation of a pipeline caused by chemical or electrochemical reactions with its environment. Corrosion is a significant concern as it ca… and cracking. Techniques for data gathering using ultrasound signals and their analysis have been optimized over decades. New generations of algorithms using artificial intelligence to detectTo sense or obtain a measurable indication from a feature. , classifyTo identify the cause of an inspection indication (e.g., anomaly, non-relevant indication, feature, or component). and size anomalies in the pipeline are replacing the previous classical approaches which have often been based on rulesets and thresholding techniques (e.g. [ 1], [2], [3], [4]). This new generation of algorithms provides the ability to
combine complex information from multiple sources to draw conclusions close to real time. This gives them high potential for the need to make critical decisions under time pressure like for priority notification where customers require fast information about highly critical anomalies in their assets.