摘要
油气管道在泄漏过程中包含大量的噪声信号,信号的提取和处理是目前泄漏检测的难题。针对淹没在无效噪声信号中的泄漏信号,提出一种基于变分模态分解(VMD)和改进的粒子群(IPSO)算法的泄漏信号处理技术。利用VMD对原始信号进行分解,以最小包络熵均值为适应度函数,采用IPSO对VMD中的分解尺度和惩罚因子进行寻优,找到最优分解效果。通过板仓-斋藤距离区分有效信号与噪声信号,提取散布熵和峭度作为特征向量,并采用支持向量机(SVM)进行工况识别。结果表明,降噪后的信号仍然保持着原始信号的走势,10 Hz以上的高频信号被消除,低频信号未受到影响,误差波形近似高斯白噪声;散布熵和峭度可以很好地表征不同工况特征,各工况特征向量无交叉,存在明显差异。该模型在测试集上的分类识别率最高,证明其在信号去噪和特征提取上的优越性。
Oil and gas pipelines contain a lot of noise signals in the process of leakage,and the extraction and processing of signals is a difficult problem in the field of leakage detection.For the leakage signal submerged in invalid noise signal,a leakage signal processing technology based on variational mode decomposition(VMD)-improved particle swarm(IPSO)algorithm is proposed.First,VMD is used to decompose the original signal,and the mean value of the minimum envelope entropy is taken as the fitness function.IPSO is used to optimize the decomposition scale and penalty factor in VMD to find the optimal decomposition effect.Then,effective signal and noise signal are distinguished by Banbana-Saito distance(ISD).Finally,distribution entropy and kurtosis are extracted as feature vectors.Support vector machine(SVM)was used to recognize the working condition.The results show that the signal after noise reduction still maintains the trend of the original signal,high frequency signal above 10 Hz is eliminated,the low frequency signal is not affected,and the error waveform is similar to Gaussian white noise.The distribution entropy and kurtosis can represent the characteristics of different conditions well,and the feature vectors of each operating condition do not cross,and there are obvious differences.The classification recognition rate of this model is the highest in the test set,which proves its superiority in signal denoising and feature extraction.
作者
韩咏梅
李昕
张军霞
万先文
刘佳庆
王渭江
HAN Yongmei;LI Xin;ZHANG Junxia;WAN Xianwen;LIU Jiaqing;WANG Weijiang(No.1 Oil Production Plant of PetroChina Huabei Oilfield Company,Renqiu,Hebei 062552,China)
出处
《世界石油工业》
2023年第3期82-89,共8页
World Petroleum Industry
关键词
油气管道
泄漏信号
变分模态分解
粒子群
散布熵
峭度
oil and gas pipeline
leakage signal
variational mode decomposition
particle swarm
dispersion entropy
kurtosis
作者简介
第一作者:韩咏梅(1975-),女,工程师,从事油气集输、自动化管理工作。E-mail:hym344605729@163.com。
