Enhanced detection orientation of buried energy pipeline damages using empirical mode decomposition and magnetic gradient tensor with detrended fluctuation analysis

Geomagnetic anomaly sensing provides a noncontact approach to assessing buried steel pipelines, yet weak defect signatures are often masked by environmental noise and uncertainty in detector orientation. This study introduces an integrated workflow that combines empirical mode decomposition with a detrended fluctuation analysis-based mode selection for adaptive denoising and employs the modulus of the magnetic gradient tensor for orientation robust localization, supported by a dipole forward model that links defect parameters to the measured field. In simulations with additive white noise (input SNR 5 dB), the EMD plus DFA pipeline achieves the strongest reconstruction among common baselines, increasing output SNR to 19.45 dB with RMSE 0.329 nT and outperforming moving average and wavelet filtering. Field trials on a Q235 steel pipeline (inner diameter about 325 mm, wall thickness 5 mm) with a 1.5 m sensor standoff and 1 kHz sampling confirm practical gains. The MGTM response forms a sharp, well-localized peak within 0.12 m of the ground truth defect, whereas single-axis gradients yield broader, less stable responses under changes in survey azimuth. The results demonstrate that coupling EMD with DFA and MGTM improves detectability and positional accuracy without excavation, using only surface measurements suitable for routine integrity management.