论文部分内容阅读
提出了一种新的基于小波包变换的信号谱峰检测算法,主要思想是利用小波包变换的特点,对信号功率谱进行平滑处理,突出谱峰的特征点(起点、顶点和终点),然后对其进行三层小波包变换,提取相应细节系数的特征点来估计谱峰的起点、顶点和终点,从而完成谱峰的检测。该方法的特点是无需信号的任何先验信息,是一种盲处理算法。仿真结果表明,信噪比不低于5dB的情况下,信号特征点检测的归一化均方误差(NMME)低于6‰,其性能比传统基于差分的方法有明显的优势。
A new detection algorithm of signal peak based on wavelet packet transform is proposed. The main idea is to utilize the characteristics of wavelet packet transform to smooth the signal power spectrum, highlight the characteristic points (starting point, vertex and end point) of the peak and then The three-layer wavelet packet transform is carried out to extract the feature points of the corresponding detail coefficients to estimate the starting point, the vertex and the ending point of the spectrum peak, thereby completing the detection of the spectrum peak. The method is characterized by a priori information without signal and is a blind processing algorithm. The simulation results show that the normalized mean square error (NMME) of signal feature point detection is less than 6 ‰ when SNR is no less than 5dB, and its performance is obviously superior to the traditional method based on difference.