论文部分内容阅读
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。
Against the background of the application of safety precaution system in oil pipelines, aiming at the problem that the traditional method to diagnose the intrusion signal collected by optical fiber is not accurate, an intrusion signal diagnosis based on empirical mode decomposition (EMD) algorithm and spectrum centroid (SC) method. Firstly, the original intrusion signal is decomposed by EMD to separate the most noisy IMFs, and then the remaining IMF components are combined to form a reconstructed signal. The reconstructed signal is subjected to Hilbert transform (HT ) To get the Hilbert spectrum, calculate its SC, further identify intrusion signals and interference signals. Through the experiment of tubing vibration signal, the diagnostic accuracy of this method is above 90.00% for each intrusion signal and interference signal, and the overall diagnostic accuracy rate reaches 97.17%. For this group of tubing vibration signals, the SVD method was used to diagnose the results and the results were compared with the diagnostic results of this method. On the whole, the diagnosis accuracy of this method is 19.00% higher than that of SVD method. The simulation results show that this method can effectively diagnose the intrusion signal, and the diagnostic results are better than the singular value decomposition.