非平稳噪声背景下瞬态声脉冲检测新方法

A new method of transient acoustic pulse detection under the nonstationary noise background

针对非平稳噪声背景下瞬态声脉冲检测难的问题,基于倒谱信号的时域分布特征,提出利用倒谱—小波进行瞬态脉冲检测的新方法,经严格的理论推导证明了该方法的可行性。数值仿真和实验数据处理表明,倒谱—小波方法的检测概率远优于传统能量检测和基于DFT的Power-law检测,尤其在极低信噪比条件下,传统检测方法皆已失效,而本文方法仍然可以获得良好的检测效果。本文工作为进一步研究水声瞬态脉冲的检测提供了技术途径和理论依据。

According to the difficulty of transient pulse detection under the nonstationary noise background, based on the distributive characteristics of cepstrum signal in time domain, this paper proposes a new way of cepstrum-wavelet to detect transient pulse. It is verified with strict theoretical calculation. Simulation and trial data processing indicate that the probability of cepstrum-wavelet outperforms conventional energy detector and Power-law detection, especially in condition of extremely low SNR, conventional methods have yet invalidated, though this method can perform well all the same. It is important as a technical approach and theoretical evidence to the research of detecting acoustic transient pulse.