Pusat Pengajian Kejuruteraan Elektrik & Elektronik - Kertas Peperiksaan / Bahan Pengajaran / Nota Latihan
Browse
Recent Submissions
Now showing 1 - 1 of 1
- PublicationNondestructive inspection of rail surface cracks using array-based magnetic flux leakage technique.(2025-02)Gong, WendongThe rapid expansion of high-speed rail networks has led to an increase in rail surface defects, making the early detection of fatigue cracks on the rail top surface crucial for preventing rail damage. Magnetic Flux Leakage (MFL) is a widely utilized technique for rail inspection, known for its speed and efficiency. However, current MFL methods often struggle to accurately estimate the severity of defects, particularly in terms of size and depth, due to the presence of outliers in the magnetic leakage field measurements. This research introduces a novel MFL detection approach that employs a Hall sensor array combined with advanced MFL signal analysis techniques to enhance inspection accuracy. First, the spatial and temporal distribution patterns of MFL signals from cracks of varying lengths, widths, depths, and orientations are analyzed using the Hall sensor array. Through orthogonal experimental analysis, the relationship between magnetic leakage flux density characteristics and crack size parameters is decoupled to establish a precise quantification sequence for crack dimensions. To further improve signal quality, the MFL signals are amplified using a magnetic field concentration sheet, and outliers caused by electromagnetic noise and vibrations are filtered via wavelet noise reduction methods. Additionally, transient geometric features of the MFL signals are extracted from the array, enabling detailed crack characterization. The crack angle is quantified using the temporal correlation of the MFL sensor array signals, while the crack width is determined through the gradient edge of the magnetic leakage flux density. The crack length is quantified by the number of sensors detecting valid MFL signals, and the crack depth is assessed based on transient geometric features. Experimental results demonstrate that the proposed method significantly improves defect detection accuracy, with the circular sensor array achieving the highest quantification accuracies. The designed circular array recorded average accuracies of 96.3%, 88.7%, 85.4%, and 93.6% for crack angle, width, length, and depth, respectively, relatively higher than square array and cross array. This study highlights the effectiveness of the proposed method in enhancing rail inspection quality and providing more accurate assessments of rail surface defects.