Implementation of a Piezo-diagnostics Approach for Damage Detection Based on PCA in a Linux-Based Embedded Platform.

Area of Science:

• Engineering
• Materials Science
• Computer Science

Background:

• Continuous structural integrity assessment requires efficient damage-detection methods.
• Embedded hardware platforms offer promising integrated solutions for Structural Health Monitoring (SHM).
• Existing methods can be computationally intensive and time-consuming.

Purpose of the Study:

• To design, test, and specify a standalone inspection prototype for structural health monitoring.
• To develop an embedded system leveraging piezo-diagnostics, Principal Component Analysis (PCA), and embedded systems.
• To demonstrate an automated, real-time diagnostic system with minimal processing resources.

Main Methods:

• Utilized a piezoelectric active system for guided wave generation and sensing.
• Embedded a Principal Component Analysis (PCA)-based algorithm on an Odroid-U3 ARM Linux platform.
• Employed a Q-statistical index (squared prediction error) to detect damages based on guided wave response deviations from a nominal PCA model.

Main Results:

• The system successfully detected induced damages, specifically added mass and leaks, in carbon-steel pipe test benches.
• Damage conditions were graphically recognized through the Q-statistic chart, correlating with experimental observations.
• Evaluated system performance on a meter carbon-steel pipe section and a pipe loop structure.

Conclusions:

• The developed embedded system is feasible for automated, real-time structural health monitoring.
• The piezo-diagnostics and PCA-based approach effectively identifies structural defects with hardware flexibility.
• The system demonstrates efficient damage detection with minimal processing requirements.