A Method for Defect Detection and Characterisation through Magnetic Flux Leakage Signals Using 3D Magnetoresistive Sensors

Health monitoring of large pipeline networks is of great importance for any country or industry. The malfunctioning of these networks may produce economic losses as well as environmental crisis. Non-Destructive Evaluation (NDE) techniques are implemented to ensure proper monitoring of pipeline networks without interfering with their operation. The Magnetic Flux Leakage (MFL) method utilises magnetic fields to detect cracks and corrosion defects on the surface of a pipe, MFL signals are then processed to characterise the detected defects. Recent research has focused on using monopolar Hall-Effect sensors to collect MFL data in order to detect regular-shaped defects. This thesis proposes a method that uses image-processing techniques for defect detection and size estimation. An experimental setup is designed in order to collect MFL signals using 3D GMR sensors and reconstruct 2D images. The approach is then replicated in a simulated environment and is tested with irregular-shaped defects in order to evaluate its accuracy with non-standard defects and attempt depth estimation by designing a novel mathematical depth function that uses the average strength of MFL signals.