经验模态分解结合频谱质心的方法在油管入侵信号诊断中的应用

Applications of empirical mode decomposition method combined with spectral centroid on the diagnosis of oil pipeline intrusion signal

在输油管道的安全防范系统应用背景下,针对传统方法诊断光纤采集到的入侵信号准确率不高的问题,提出一种基于经验模态分解(EMD)算法和频谱质心(SC)的入侵信号诊断方法。首先将采集到的原始入侵信号通过EMD进行分解,分离含噪最多的特征模态函数(IMF)分量,再组合剩余的IMF分量形成重构信号,对重构信号进行希尔伯特变换(HT)得到希尔伯特谱,计算它的SC,进一步识别入侵信号和干扰信号。通过对油管振动信号进行实验,本文方法对于每种入侵信号和干扰信号的诊断准确率均在90.00%以上,整体的诊断准确率达到97.17%。对于该组油管振动信号,同时运用奇异值分解(SVD)法进行诊断并将其结果与本文方法的诊断结果进行对比,整体上本文方法的诊断准确率比SVD法高出19.00%。仿真实验结果表明,本文方法能有效诊断入侵信号,并且诊断效果明显优于奇异值分解法。

Under the background of the application of the tubing safety guard system, an intrusion signal diagnosis method based on empirical mode decomposition （EMD） algorithm and spectrum centroid （SC） is introduced,in view of the problems of low diagnostic accuracy of intrusion signals collected by optical fiber using the conventional method. Firstly,this method decomposes the original intrusion signal through EMD, and separates the intrinsic mode function （IMF） component with the most noise. Then, the recon- structed signal is formed by the combination of the remaining IMF components, and the Hilbert spectrum is obtained by Hilbert transform of the reconstructed signal. Next, its SC is calculated. Finally, according to the threshold value method, the intrusion signals and the interference signals are diagnosed. The re- sults of the simulation experiment of the tubing vibration signals show that the diagnostic accuracy of each kind of intrusion signal and interference signal by the proposed method is more than 90.00% ,and the overall diagnostic accuracy is 97. 17%. Compared with the results diagnosed simultaneously by the singular value decomposition （SVD） method, a higher diagnostic accuracy of 19.00 % for the proposed method is obtained on the whole in the former setting. The results of the simulation experiment show that the proposed method can effectively diagnose the intrusion signals, and its diagnostic effect is obvi-ously superior to that of the SVD method.