Research on a Method for Classifying Bolt Corrosion Based on an Acoustic Emission Sensor System.

Area of Science:

• Structural Health Monitoring
• Materials Science
• Signal Processing

Background:

• High-strength bolts are critical components in infrastructure like bridges and railways.
• Monitoring bolt integrity is essential for preventing accidents and facilitating repairs.
• Corrosion significantly degrades bolt performance and poses safety risks.

Purpose of the Study:

• To develop and validate a novel system for accurate detection and classification of bolt corrosion levels.
• To enhance the reliability and safety of ultra-high-pressure equipment through effective bolt condition monitoring.
• To establish a robust method for assessing bolt degradation using acoustic emission signals.

Main Methods:

• Implementation of a Wireless Acoustic Emission Sensor Network (WASN) for high-speed signal acquisition.
• Application of the ReliefF algorithm for optimal feature selection from acoustic emission data.
• Utilization of the Extreme Learning Machine (ELM) model for corrosion classification.
• Optimization of ELM parameters using an improved goose algorithm (GOOSE) for enhanced prediction accuracy.

Main Results:

• Experimental validation across five distinct bolt corrosion levels (0% to 100%).
• Achieved a minimum classification accuracy of 98.04% for detecting bolt corrosion.
• Demonstrated superior performance compared to existing classification diagnostic models.
• Showcased strong generalization ability across varying working conditions.

Conclusions:

• The proposed WASN-based system offers a highly accurate and reliable method for bolt corrosion classification.
• The integration of ReliefF and GOOSE-optimized ELM provides robust AE signal recognition.
• This approach significantly contributes to the structural health monitoring and safety of critical infrastructure.