针对在混合噪声环境中,在未知噪声先验信息的情况下提高基于广义互相关时延估计方法的准确性和适应性的问题,提出了基于神经网络滤波的广义互相关时延估计方法.该方法通过对多个具有特定统计特征的预滤波器进行组合优化来实现对混合噪声的滤波.该时延估计方法具有自调整能力,能够适应动态变化的环境,还能够通过相关函数峰值的大小来优化神经网络滤波器,提高了适应性和精度.%In the mixed noise environment where the priori information of noise is unknown, the in order to improve both accuracy and adaptability of the method based on generalized cross-correlation time-delay estimation, an estimation method based on neural network filtering was proposed to have the mixed noise filtered by grouping and optimizing multiple pre-filters with specific statistical characteristics. This method with self-adjustment , adapt to dynamic environments and able to optimize neural network filter through the peak size of cross-correlation function can improve both adaptability and accuracy.
展开▼