基于特征量提取的输气管道微泄漏检测

输气管道音波法泄漏检测采集得到的信号不仅包含有用的泄漏信号,而且包含背景噪声和各种干扰信号,因此信号的识别和特征量提取尤为重要。基于此情况,采用相关性分析的方法对传感器采集的信号进行处理并得到有效特征量,传感器采集得到的信号包括泄漏信号、敲击信号、压缩机启停信号、减压阀开关信号。相关性分析采用相关函数和协方差函数实现,相关函数可以得到各种信号的自相关和互相关特征,协方差函数可以得到各种信号的自协方差和互协方差特征。同时对信号进行整体峰度计算,并设置整体峰度阈值。研究结果表明:信号的相关性分析可以对输气管道微泄漏进行检测,同时对诸如减压阀操作、压缩机启停、敲击等干扰因素可以通过相关函数数值从背景噪声中识别;在不确定是否存在干扰信号的前提下,通过相关分析从背景噪声中提取泄漏信号或干扰信号,并对信号进行整体峰度值计算,若整体峰度值高于阈值,则认为泄漏发生,提高了音波泄漏检测的准确性。

Characteristics extraction of acoustic leakage signal for natural gas pipelines

The signals acquired by acoustic leak detection method for natural gas pipelines include useful leakage signal, background noises and interference signals. Therefore, the recognition and characteristics extraction of leakage signal are getting more and more important. The measured signals including leakage signal, knocking signal, compressor shutoff signal, compressor starting signal, regulator closing signal and regulator openning signal were processed by correlation analyses method based on correlation and covariance function. The characteristics of auto-correlation and cross-correlation were extracted. The results show that the correlation analyses can extract the characteristics of leakage signals and other interference signals from the background noises. Then the overall kurtosis can be applied to distinguish the leakage signals from the interference signals. The characteristics extracted by correlation analyses are effective for recognizing leakage signal among the noises and the characteristics of the overall kurtosis can be used to detect leakage signals among all acquired signals if the threshold value of the overall kurtosis is set, which has a strong impetus to the improvement and application of acoustic leak detection technology.