Metal fatigue cracks under cyclic loading conditions cannot be ignored. Ultrasound-induced thermography has been emerged as an effective technique for crack detection, however, single-frequency ultrasound excitation often produces weak thermal signals, especially for small or shallow cracks, limiting identification accuracy. To overcome these challenges, Multimodal Algorithm-enhanced Chirped Ultrasound-Induced Thermography (MA-CUIT) was proposed for quantitative crack characterization. Firstly, metal cracks three-dimensional thermal-wave model under chirp excitation was established to analyze temperature distribution and thermal diffusion. Subsequently, a comprehensive feature extraction framework integrating frequency-domain, time-domain, and machine learning methods was developed under chirp ultrasound excitation, with systematic optimization of frequency modulation parameters to determine optimal excitation ranges. MA-CUIT experimental platform was established for quantification of real crack defects. The results indicate that MA-CUIT, achieved exceptional crack sizing accuracy with relative errors of 5.9% (FFT phase optimized), 6.7% (PLSR 1st coefficient optimized), and 6.3% (DOD phase optimized), establishing MA-CUIT optimized by FFT phase analysis as the most precise quantification method.

Multimodal algorithm-optimized chirped ultrasound thermography for enhanced quantitative detection of metal fatigue cracks

Sfarra, Stefano;
2026-01-01

Abstract

Metal fatigue cracks under cyclic loading conditions cannot be ignored. Ultrasound-induced thermography has been emerged as an effective technique for crack detection, however, single-frequency ultrasound excitation often produces weak thermal signals, especially for small or shallow cracks, limiting identification accuracy. To overcome these challenges, Multimodal Algorithm-enhanced Chirped Ultrasound-Induced Thermography (MA-CUIT) was proposed for quantitative crack characterization. Firstly, metal cracks three-dimensional thermal-wave model under chirp excitation was established to analyze temperature distribution and thermal diffusion. Subsequently, a comprehensive feature extraction framework integrating frequency-domain, time-domain, and machine learning methods was developed under chirp ultrasound excitation, with systematic optimization of frequency modulation parameters to determine optimal excitation ranges. MA-CUIT experimental platform was established for quantification of real crack defects. The results indicate that MA-CUIT, achieved exceptional crack sizing accuracy with relative errors of 5.9% (FFT phase optimized), 6.7% (PLSR 1st coefficient optimized), and 6.3% (DOD phase optimized), establishing MA-CUIT optimized by FFT phase analysis as the most precise quantification method.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11697/290370
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 17
  • ???jsp.display-item.citation.isi??? 15
social impact