Optical Rail Surface Crack Detection Method Based on Semantic Segmentation Replacement for Magnetic Particle...

Area of Science:

• Engineering
• Materials Science
• Computer Science

Background:

• Railway safety relies heavily on detecting rail surface cracks, which are crucial for understanding damage progression and predicting defects.
• Traditional crack detection methods are often slow, complex, and costly, hindering efficient railway maintenance.

Purpose of the Study:

• To develop an efficient and accurate semantic segmentation method for detecting rail surface cracks using deep learning.
• To provide an economical and fast alternative to conventional rail inspection techniques.

Main Methods:

• A deep learning approach utilizing a neural network was employed for semantic segmentation of rail surface crack data.
• The performance of the deep learning method was compared against traditional magnetic particle inspection technology.

Main Results:

• The semantic segmentation method achieved high accuracy in detecting rail surface cracks.
• The camera-based deep learning approach significantly increased inspection speed compared to magnetic particle inspection.
• The method demonstrated potential for even faster detection when integrated with high-frequency cameras.

Conclusions:

• The developed deep learning method offers an economical, efficient, and environmentally friendly solution for rail surface crack detection.
• This innovative approach enhances railway safety by providing a faster and more accessible inspection technique.