中文摘要：

针对声发射管道泄漏检测过程中的噪声干扰问题，对基于小波包和经验模态分解（EMD）的声发射信号处理方法进行了研究。采用小波包分解算法和经验模态分解都可以对管道泄漏声发射信号进行分解，但分解结果却存在一定区别。EMD是近年来非平稳信号分析领域的一个突破，对管道泄漏声发射信号进行EMD分解后，选择包含声发射特征的若干固有模式函数（IMF分量）进行重构，可以提取到管道泄漏声发射信号的本质特征，消除噪声信号的干扰。相对小波包分解方法而言，对根据IMF分量重构的声发射信号进行相关分析计算，得到的管道泄漏点的位置更为精确。

英文摘要：

Aiming at the noise interference problems occurred in acoustic emission pipeline leakage detection, signal analysis method based on wavelet package and empirical mode decomposition （EMD） were researched, respectively. Both wavelet package and EMD algorithms can be used to decompose pipeline leakage AE signals, but the results are greatly different. EMD is a great breakthrough in non-stable signal analysis. Using EMD the detected leakage signal can be decomposed into a sum of finite intrinsic mode functions （ IMF）, among which some IMF components contai- ning typical AE characteristic can be selected to reconstruct the signal, and thus intrinsic characteristic of the leak- age signal can be extracted and noise interference can be eliminated. Comparing with wavelet package algorithm, the location accuracy of the leaking point calculated from the reconstructed signals based on EMD is greatly increased.