Efficient Near-Field Radiofrequency Imaging of Impact Damage on CFRP Materials with Learning-Based Compressed Sensing

Area of Science:

• Materials Science
• Non-Destructive Testing
• Composite Materials

Background:

• Carbon fiber-reinforced polymers (CFRP) are susceptible to internal impact damage.
• Near-field radiofrequency imaging (NRI) detects subsurface defects but is time-consuming.
• Existing compressed sensing (CS) methods for NRI reduce measurement time but have high reconstruction times.

Purpose of the Study:

• To develop a deep learning-based compressed sensing (CS) method for accelerating near-field radiofrequency imaging (NRI).
• To enable real-time imaging and reduce data acquisition for CFRP defect detection.
• To improve the efficiency of NRI without requiring hardware modifications.

Main Methods:

• A novel 0/1-Bernoulli measurement matrix was designed for sensor scanning.
• An interpretable neural network-based CS reconstruction method was developed.
• The deep learning approach was integrated into the NRI system for accelerated data acquisition and reconstruction.

Main Results:

• The proposed method decreased NRI measurement time by one order of magnitude.
• Real-time imaging was achieved during CS reconstruction.
• Experimental results demonstrated a 20x speed increase and over 90% data reduction compared to traditional methods and existing CS techniques, maintaining imaging quality.

Conclusions:

• Deep learning-based CS significantly accelerates NRI for CFRP inspection.
• The method offers a practical solution for real-time, efficient subsurface defect detection.
• This approach enhances the applicability of NRI in composite material analysis and maintenance.