Improved measurement of the glue layer in composite material by using sparse deconvolution

Abstract

Due to its powerful penetration, and greater spatial resolution than microwaves and ultrasonic waves, the terahertz technique stands out as being particularly useful in identifying thin glue layers in multilayered materials. However, the arrival times of echoes are challenging to pinpoint from the experimental data because of the temporal form of the incident pulse and the system noise. Here, two terahertz signal sparse deconvolution algorithms are studied to more accurately identify the times of the echoes. Using the circulant structure of the convolution matrix, the method’s computation time can be lowered to hundreds of milliseconds. In addition, a method based on group velocity dispersion is investigated to reduce the impact of time-varying pulses with minimal computational expense. The presented algorithms have the potential to be employed in real-time inspection in production lines due to their quick speed and high confidence.