Automatic Detection of Real Damage in Operating Tie-Rods.

Area of Science:

• Civil Engineering
• Mechanical Engineering
• Materials Science

Background:

• Vibration-based damage detection for structural health monitoring (SHM) faces challenges in real-world applications due to uncontrolled environmental and operational variations.
• Existing methods often fail to detect actual damage in realistic scenarios, limiting their practical implementation.
• Previous research primarily validated damage detection algorithms in controlled settings with simulated damage.

Purpose of the Study:

• To develop a fully-automated, single-sensor strategy for detecting damage in operating tie-rods.
• To enable damage detection without requiring knowledge of physical variables like axial load.
• To enhance the reliability of damage detection in real-world, uncontrolled environments.

Main Methods:

• Defined a damage feature based on tie-rod eigenfrequencies.
• Developed a data-cleansing strategy to improve outlier detection performance using Mahalanobis squared distance.
• Validated the approach using real damage from corrosion in an uncontrolled environment.

Main Results:

• The proposed strategy successfully detected real tie-rod damage caused by corrosion.
• The method demonstrated effectiveness even under intentionally uncontrolled environmental conditions.
• The data-cleansing strategy significantly improved the performance of outlier detection.

Conclusions:

• The developed automated strategy offers a robust solution for early detection of real tie-rod damage.
• This approach overcomes limitations of previous methods by performing effectively in uncontrolled environments.
• The single-sensor, load-independent method enhances the practicality of vibration-based SHM for operating structures.