CAJ | 학술논문

提出一种基于局域均值分解的RMS熵及时延估计的天然气管道泄漏孔径识别及定位方法。该方法首先对泄漏信号进行 LMD 分解，得到若干具有物理意义的 PF 分量并计算其有效值，进而结合信息熵的概念得出不同泄漏孔径的RMS 熵，将不同孔径泄漏信号的多个 RMS 熵组成特征向量输入 SVM 进行识别。为提高互相关法定位精度，提出根据LMD分解结果的峭度特征进行重构再进行互相关获取时延信息，并结合泄漏信号的传播速度，实现泄漏点定位。实验结果表明该方法能够实现管道泄漏孔径有效识别及定位，且与基于 EMD 的 RMS熵方法相比，识别效果更好，较直接相关法的定位精度明显提高。
제출일충기우국역균치분해적RMS적급시연고계적천연기관도설루공경식별급정위방법。해방법수선대설루신호진행 LMD 분해，득도약간구유물리의의적 PF 분량병계산기유효치，진이결합신식적적개념득출불동설루공경적RMS 적，장불동공경설루신호적다개 RMS 적조성특정향량수입 SVM 진행식별。위제고호상관법정위정도，제출근거LMD분해결과적초도특정진행중구재진행호상관획취시연신식，병결합설루신호적전파속도，실현설루점정위。실험결과표명해방법능구실현관도설루공경유효식별급정위，차여기우 EMD 적 RMS적방법상비，식별효과경호，교직접상관법적정위정도명현제고。
Aiming at natural gas pipeline leak problem, a leak apertures recognition and location method based on RMS (root mean square) entropy and time delay estimation is presented by analyzing local mean decomposition (LMD) results. The leak signals are decomposed by LMD and several PF (product function) components with clearly physical meaning are obtained. The PF components RMS is calculated, which combines information entropy to acquire RMS entropy ofdifferent leak apertures. Several RMS entropy values are chosen as the feature vectors and input to the SVM to achieve the identification. In order to improve the location accuracy of cross-correlation, the kurtosis of PF components is analyzed, and leak signals are constructed based on principal PFs to improve time delay estimation accuracy. Combining the stress wave velocity, the leak location is caccomplished. Experimental results show the proposed methods with LMD analysis can effectively identify apertures and locate the leak, and the recognition result is better than the RMS entropy based on EMD. Location accuracy is obviously improved than the direct cross-correlation method.