基于EWT及模糊相关分类器的管道微小泄漏检测

Pipeline small leakage detection based on the EWT and AFCC

针对管道微小泄漏难以检测的问题,提出一种基于经验小波变换(Empirical Wavelet Transform,EWT)及模糊相关分类器的检测方法。该方法首先对传感器采集的信号进行EWT分解,计算分解后分量的峭度值,根据分量的峭度值提出基于EWT的自适应降噪法对信号进行降噪处理,提取无噪声分量并重构;分析重构信号的模糊函数图像,结合相关系数提出模糊相关分类器检测管道微小泄漏。仿真及实验结果表明,该方法能够对管道微小泄漏进行检测,且检测率较支持向量机及神经网络高。

Aiming at the small leakage detection in natural gas pipelines,a leak detection method based on the empirical wavelet transform（ EWT） and ambiguity function correlation classification（ AFCC） was proposed. The acquired signals from sensors were decomposed by the EWT,and then the kurtosis values of decomposition components were calculated. According to the kurtosis values,an adaptive de-noising algorithm based on the EWT was proposed to process these noised components and the de-noised components were extracted and then reconstructed. The ambiguity function images were employed to analyze the de-noised signals. Combining with the correlation coefficient,the AFCC was proposed to detect the small leakage of pipelines. The experimental results show the proposed method can effectively detect the small leakage,and the result is better than that of the SVM or BP.