Quantitative detection for textured surface cracks based on cluster integration.

Area of Science:

• Materials Science
• Nuclear Engineering
• Image Processing

Background:

• Zirconium plates are critical nuclear reactor components.
• Manufacturing defects, such as surface cracks, can lead to component failure and nuclear fuel leakage.
• Accurate crack detection is essential for ensuring nuclear safety and integrity.

Purpose of the Study:

• To develop a precise quantitative detection scheme for cracks on zirconium plates.
• To address the challenges of detecting diverse crack patterns and intensive horizontal line structures.
• To improve the accuracy and stability of crack measurement.

Main Methods:

• Proposed a quantitative detection scheme incorporating a two-phase cluster integration strategy (TPCIS).
• Utilized trunk reconstruction measurement evaluation for precise defect area segmentation.
• Employed artificial cracks as a dataset for algorithm performance analysis.

Main Results:

• The TPCIS achieved an average crack detection rate of 92.9%.
• Relative errors for measuring crack area, length, and width (down to 50 µm) were controlled within 1.5%.
• Demonstrated higher accuracy and stability compared to previous crack detection methods.

Conclusions:

• The developed TPCIS offers a robust solution for precise crack detection and quantitative morphological analysis on zirconium plates.
• The proposed method significantly enhances the reliability of zirconium plate inspection in nuclear applications.
• This advancement contributes to improved nuclear fuel safety and reactor system integrity.