基于噪声预估计的DPOAE最小二乘估计方法*

针对畸变产物耳声发射(distortion product otoacoustic emission, DPOAE)信号比较微弱,估计结果的准确性对信噪比依赖性强的特点,本文给出一种新的DPOAE信号最小二乘估计(least square estimation, LSE)方法。该方法先估计本底噪声,将本底噪声的影响体现于估计模型中,再估计DPOAE信号。记录信号经过一定的预处理后,利用频域加窗法估计本底噪声并进行噪声衰减；利用IFFT变换后的信号估计DPOAE信号的模型参数,给出DPOAE的强度和相位。仿真和实验结果表明,该方法相对于传统的最小二乘拟合(least-square-fit, LSF)法,对DPOAE的估计更加准确,抗噪声干扰能力更强。

Least square estimation of distortion product otoacoustic emission signals based on noise pre-estimation

Since distortion product otoacoustic emission signal (DPOAE) is relatively weak and the estimation result strongly depends on the signal-to-noise ratio, this paper presents a new DPOAE estimation method, least square estimation method, in which noise floor is introduced. The noise floor has been estimated and the impact of background noise is considered in the estimated model. After pretreatment, the background noise has been estimated based on the windowed FFT spectrum and a certain degree of noise attenuation has been applied, followed by the IFFT transform to obtain time-domain signal. DPOAE has been modeled and model parameters have been estimated using time-domain data. Ultimately, the intensity and phase of the DPOAE signal are calculated based on the estimated model parameters. Simulation and experimental results show that more accurate estimate of the DPOAE signal is obtained by the method relative to traditional least-square-fit because of its stronger anti-interference ability.