Non-destructive Detection of Pipeline Properties and Stress Using Geostatistical Pattern Recognition Analysis of In-Line Inspection Magnetic Flux Leakage Data
This project will evaluate and apply novel spatial pattern recognition methods (neutral models) for use in the analysis of In-line Inspection Magnetic Flux Leakage (ILI MFL) data to detect signals symptomatic of pipeline stress, bending strain and other pipe properties such as wall thickness and pipe strength. Neutral models are data mining techniques that allow flexible specification of “background”, where background consists of those spatial patterns that are not considered to be of interest.
For ILI MFL data, background will be specified as those signals expected from pipes within acceptable operating environments. The promise for pipeline O&I applications is the accurate in-line detection of pipeline stress and strain prior to metal deformation and ablation – signals that cannot be found using existing state-of-practice analysis techniques.