Denoising low SNR percussion acoustic signal in the marine environment based on the LMS algorithm. (October 2022)

• Record Type: Journal Article
• Authors: Yang, Zhuodong; Huo, Linsheng; Wang, Jingkai; Zhou, Jing
• Is Part Of: Measurement. Volume 202(2022)
• Publication Date: 2022-10
• Subjects: LMS algorithm -- Marine noise -- Percussion acoustic signal -- Signal denoising; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2022.111848
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: A percussion acoustic signal denoising method in marine environment is proposed. Compared with other denoising methods, the proposed method can significantly improve the SNR of noisy percussion acoustic signals under the complex marine environment with noise interference. When the main peak frequency of the denoised percussion acoustic signal with the proposed method is used for damage identification, the error is only about 3% Abstract: Percussion-based inspection of structures has attracted widespread attention in recent years. However, the percussion acoustic signals collected in the marine environment usually have a low signal-to-noise ratio (SNR) and are difficult to use directly due to the interference by a multitude of marine noises. The frequency contents of the ambient noises usually overlap with those of the percussion acoustic signals, thus limiting the denoising using traditional methods. This paper proposes a denoising method using the least mean square (LMS) algorithm to obtain the approximate percussion signal. The noisy percussion signals and marine noise are recorded synchronously by two hydrophones. Then the LMS algorithm processes the collected signals and provides the frequency peaks that cannot be extracted with conventional methods. The proposed method is validated by experiments conducted in a noiseless laboratory environment and a noisy, naturally occurring marine environment. The results reveal that the proposed method is excellent in denoising the raw signal, and the error is about 3% in terms of differences in the estimated value of the primary peak frequency. This study demonstrates the broad potential for the method to be applied toward damage detection for underwater structures.