Features of Cross-Correlation Analysis in a Data-Driven Approach for Structural Damage Assessment.

Area of Science:

• Engineering
• Materials Science
• Data Science

Background:

• Structural Health Monitoring (SHM) relies on accurate data analysis for damage detection.
• Traditional methods can struggle with noisy data and require robust preprocessing.
• Piezodiagnostics offers a promising approach for SHM using piezoelectric devices.

Purpose of the Study:

• To evaluate the advantage of cross-correlation analysis in a data-driven SHM approach.
• To demonstrate improved damage detection through preprocessing with cross-correlation functions.
• To validate a methodology combining cross-correlation, principal component analysis (PCA), and piezodiagnostics.

Main Methods:

• Implementation of a preprocessing stage using cross-correlation functions for data cleansing.
• Application of principal component analysis (PCA) for data-driven analysis.
• Validation using data from piezoelectric (PZT) devices on laboratory specimens (pipe, aircraft wing, turbine blade).

Main Results:

• Cross-correlation analysis facilitates identification of noisy data and outliers.
• Inclusion of cross-correlation in the preprocessing stage significantly improves damage detection accuracy.
• The methodology effectively detected various damage types, including leaks and mass additions.

Conclusions:

• Cross-correlation analysis is a valuable tool for enhancing data quality in SHM.
• Combining cross-correlation features with PCA-based piezodiagnostics yields a more robust damage assessment algorithm.
• The validated methodology shows significant potential for practical SHM applications.