Researchers from Southwest Jiaotong University and The Hong Kong Polytechnic University have developed an automated fatigue crack detection system for orthotropic steel bridge decks (OSDs), which are critical to long-span bridge designs.

Essential for their high load-carrying efficiency and lightweight design, OSDs incorporate intricate structures that are particularly susceptible to fatigue cracking at critical connection points, creating significant safety concerns.

Source: Journal of Infrastructure Intelligence and Resilience (2024). DOI: 10.1016/j.iintel.2024.100113

Common inspection methods — such as visual checks and magnetic testing — tend to lack the precision and reliability necessary for detecting internal or subtle cracks within OSDs. Meanwhile, Phased Array Ultrasonic Testing (PAUT) has shown promise for this application, but issues like those noted previously persist.

As such, the researchers have introduced its automated system for fatigue crack detection in OSDs, which uses a robotic platform in conjunction with ultrasonic phased array technology.

This new system is reportedly enhanced by deep learning models including Deep Convolutional Generative Adversarial Network (DCGAN) for data generation and YOLOv7-tiny for high-speed, real-time crack detection. This approach thus enables a dramatic improvement in both accuracy and efficiency, thereby promising to revolutionize future bridge maintenance practices.

The team explained that the robotic system is equipped with a phased array ultrasonic probe that can autonomously scan OSDs, which reduces the need for human involvement.

The researchers explained: "Our research addresses key safety concerns in bridge maintenance by harnessing robotic automation and deep learning technologies. The result is a highly efficient system that can detect fatigue cracks with unprecedented accuracy, even in challenging conditions.

The team noted that by automating the inspection of OSDs, the need for manual labor is significantly reduced. Likewise, it enables early detection of structural issues, thereby preventing catastrophic failures.

The article, "Automatic PAUT crack detection and depth identification framework based on inspection robot and deep learning method,” appears in the Journal of Infrastructure Intelligence and Resilience.