Parameterizing Magnetic Flux Leakage Data for Pipeline Corrosion Defect Retrieval

Magnetic flux leakage (MFL) is the most popular in-line inspection (ILI) technique to inspect pipeline corrosion. The collected MFL signals are characterized to estimate the profile of corrosion defects. However, the estimation error could be huge for certain corrosion areas because of the signal interference between adjacent defects. To retrieve these corrosion areas from the whole pipeline, one accurate and reliable representation of the corrosion defect is critical while no relevant research has been done yet. In this study, the concept of MFL data parameterization is proposed first. Parameterization is a contextual defect representation, which considers both corrosion defect and its surroundings to deal with the signal interference. Besides, one two-dimensional Gaussian function is introduced to denote the interference strength, and three parameterization models are then developed to obtain a reliable representation of corrosion defect. In the end, two experiments on corrosion defect retrieval are conducted to evaluate the performance of three parameterization models.