论文部分内容阅读
提出一种基于112维谱估计的直升机声信号低频线谱特征提取方法。112维谱是高阶谱的一个特例,它具有几个特殊的性质,这些性质使其能抑制高斯噪声、加强谐波信号基频分量以及剔除非二次相位耦合的谐波分量,因此在处理线谱估计问题时112维谱估计要比传统的功率谱估计方法具有更好的性能;同时与双谱估计等常用的高阶谱估计方法相比,112维谱估计大大简化了计算。通过对实测直升机噪声数据进行仿真,结果证明,将112维谱估计用于直升机声信号的特征提取可行、有效。
This paper presents a low-frequency spectral feature extraction method of helicopter acoustic signal based on 112-dimensional spectrum estimation. As a special case of high-order spectra, the dimensionality spectrum 112 has several special properties that make it possible to suppress Gaussian noise, enhance the fundamental frequency components of harmonic signals, and reject the harmonic components of non-quadratic phase couplings. Therefore, Compared with the traditional power spectrum estimation methods, the 112-dimensional spectrum estimation has better performance than the traditional spectral estimation methods. Compared with the commonly used high-order spectral estimation methods such as bispectrum estimation, the 112-dimensional spectral estimation greatly simplifies the calculation. Through the simulation of measured helicopter noise data, the results show that it is feasible and effective to use 112 dimension spectrum estimation in the helicopter acoustic signal feature extraction.