Digital signal processing has become an important component of data analysis needed in industrial applications. In particular, for ultrasonic thickness measurements, the signal-to-noise ratio (SNR) plays a major role in the accurate calculation of the arrival time. For this application, a bandpass filter is not sufficient since the noise level cannot be significantly decreased such that a reliable thickness measurement can be performed. This paper demonstrates the abilities of two regularization methods - total variation (TV) and Tikhonov - to filter acoustic and ultrasonic signals. Both of these methods are compared with frequency-based filtering for digitally produced signals as well as signals produced by ultrasonic transducers. This paper demonstrates the ability of the TV and Tikhonov filters to accurately recover signals from noisy acoustic signals faster than a bandpass filter. Furthermore, the TV filter has been shown to reduce the noise of a signal significantly for the signals with clear ultrasonic echoes. SNRs have been increased over 400% by using simple parameter optimization. While frequency-based filtering is efficient for specific applications, this paper shows that the reduction of noise in ultrasonic systems can be much more efficient with regularization methods.
|Original language||English (US)|
|Number of pages||5|
|Journal||IEEE transactions on ultrasonics, ferroelectrics, and frequency control|
|State||Published - Aug 1 2016|
All Science Journal Classification (ASJC) codes
- Acoustics and Ultrasonics
- Electrical and Electronic Engineering