Advancing Data Completeness And Record Reviews With Machine Learning

Paper No. IPC2024-134143

This study introduces a machine learning approach to predict pipeline stress as a percentage of the Specified Minimum Yield Strength (% SMYS), addressing data gaps. The models help classify segments for integrity management, reducing manual reviews. By identifying high-stress segments, especially those over 30% SMYS, and using models like random forest and XGBoost, the approach improves accuracy with feedback from manual reviews, demonstrating machine learning's potential in pipeline assessments.