基于小波包能量及高阶谱的特征提取方法
|
摘要
针对广泛存在的油气管道周边安全问题,研究了管道周围地面活动目标产生的震动信号的特性,提出了一种基于小波包能量谱和信号高阶谱分析相结合的特征提取方法来区分不同的活动目标.根据目标产生的地面震动信号是非平稳的特点,采用基于小波包分解能量的方法对信号的各频带进行分解,得到信号在不同频带内的能量分布特性.仅根据能量谱并不能完全区分不同类型信号,通过对信号高阶统计特性的分析,提取出高阶谱特征频率,结合这两种方法提取出的特征作为神经网络的输入向量进行模式识别.通过对实验数据进行分析,单独采用小波包能量特征其平均识别率为88.5%,而采用本文提出的方法平均识别率可以提高到94.6%,验证了文中提出方法的有效性.
In view of the prevalent peripheral security problems around oil and gas pipelines,characteristics of seismic signals generated by moving targets on the surrounding ground have been investigated and a feature extraction method based on wavelet packet energy distribution and high-order spectrum analysis has been put forward to distinguish the different targets.As the seismic signals generated by ground targets were not stable,wavelet packet energy method was used to decompose the signals at several independent frequency bands and to draw the energy distribution features in these bands.However,the wavelet packet energy distribution could not distinguish the types of signals very well.Therefore,high-order statistic features of signals were analyzed and the characteristic frequencies were extracted.The features extracted based on wavelet packet energy distribution and high-order spectrum were input into the neural network as eigenvectors for pattern recognition.Experiment results indicate that the average recognition rate is 88.5% with only wavelet packet energy spectrum based method while it increases to 94.6% with the method proposed,thus verifying its effectiveness.
In view of the prevalent peripheral security problems around oil and gas pipelines,characteristics of seismic signals generated by moving targets on the surrounding ground have been investigated and a feature extraction method based on wavelet packet energy distribution and high-order spectrum analysis has been put forward to distinguish the different targets.As the seismic signals generated by ground targets were not stable,wavelet packet energy method was used to decompose the signals at several independent frequency bands and to draw the energy distribution features in these bands.However,the wavelet packet energy distribution could not distinguish the types of signals very well.Therefore,high-order statistic features of signals were analyzed and the characteristic frequencies were extracted.The features extracted based on wavelet packet energy distribution and high-order spectrum were input into the neural network as eigenvectors for pattern recognition.Experiment results indicate that the average recognition rate is 88.5% with only wavelet packet energy spectrum based method while it increases to 94.6% with the method proposed,thus verifying its effectiveness.
引文
[1]靳世久,唐秀家,王立宁,等.原油管道泄漏检测与定位[J].仪器仪表学报,1997,18(4):343-348.Jin Shijiu,Tang Xiujia,Wang Lining,et al.Detection and location of leakage in crude petroleum pipeline[J].Chinese Journal of Scientific Instru-ment,1997,18(4):343-348(in Chinese).
[2]崔谦.油气管道泄漏检测方法的研究及应用[D].天津:天津大学精密仪器与光电子工程学院,2005.Cui Qian.Study on Oil and Gas Pipeline Leak Detection Methods and Their Application[D].Tianjin:School of Precision Instrument and Opto-Electronics Engineering,Tianjin University,2005(in Chinese).
[3]陈仁文.小波变换在输油管道漏油实时监测中的应用[J].仪器仪表学报,2005,26(3):242-245.Chen Renwen.The applications of wavelet transform in oil leakage online detection[J].Chinese Journal of Sci-entific Instrument,2005,26(3):242-245(in Chinese).
[4]Pakhomov A,Goldburt T.New seismic unattended small module for foot-step detection[C]//Proc of SPIE:Unattended Ground,Sea,and Air Sensor Technologies and Applications VIII,_United States:SPIE.Orlando,FL,USA,2006,6231:623108.
[5]姚姚.地震波场与地震勘探[M].北京:地质出版社,2006:22-26.Yao Yao.Seismic Wave Field and Seismic Exploration[M].Beijin:Geological Publishing House,2006:22-26(in Chinese).
[6]贺银芝,沈松,应怀樵,等.小波包分解及其能量谱在发动机连杆轴承故障诊断中的应用[J].振动工程学报,2001,14(1):72-75.He Yinzhi,Shen Song,Ying Huaiqiao,et al.Application of wavelet packet decomposition and its energy spectrum on the fault diagnosis of reciprocation machinery[J].Journal of Vibration Engineering,2001,14(1):72-75(in Chinese).
[7]丁幼亮,李爱群,缪长青.基于小波包能量谱的结构损伤预警方法研究[J].工程力学,2006,23(8):29-38.Ding Youliang,Li Aiqun,Miu Changqing.Investigation on the structural damage alarming method based on wavelet packet energy spectrum[J].Engineering Me-chanics,2006,23(8):29-38(in Chinese).
[8]郑海波,陈心昭,李志远.基于双谱的齿轮故障特征提取与识别[J].振动工程学报,2002,15(4):354-358.Zheng Haibo,Chen Xinzhao,Li Zhiyuan.Bispectrum based gear fault feature extraction and diagnosis[J].Journal of Vibration Engineering,2002,15(4):354-358(in Chinese).
[9]Zhang Guicai,Chen Jin,Li Fucai,et al.Extracting gear fault features using maximal bispectrum[J].Key Engi-neering Materials,2005,293/294:167-174.
[10]曲志刚,靳世久,周琰,等.基于RBF网络的油气管道侵入事件识别方法研究[J].化工自动化及仪表,2007,34(3):58-61.Qu Zhigang,Jin Shijiu,Zhou Yan,et al.Study on recognition method based on RBF neural network for intrusion events of oil and gas pipeline[J].Control and Instruments in Chemical Industry,2007,34(3):58-61(in Chinese).
[2]崔谦.油气管道泄漏检测方法的研究及应用[D].天津:天津大学精密仪器与光电子工程学院,2005.Cui Qian.Study on Oil and Gas Pipeline Leak Detection Methods and Their Application[D].Tianjin:School of Precision Instrument and Opto-Electronics Engineering,Tianjin University,2005(in Chinese).
[3]陈仁文.小波变换在输油管道漏油实时监测中的应用[J].仪器仪表学报,2005,26(3):242-245.Chen Renwen.The applications of wavelet transform in oil leakage online detection[J].Chinese Journal of Sci-entific Instrument,2005,26(3):242-245(in Chinese).
[4]Pakhomov A,Goldburt T.New seismic unattended small module for foot-step detection[C]//Proc of SPIE:Unattended Ground,Sea,and Air Sensor Technologies and Applications VIII,_United States:SPIE.Orlando,FL,USA,2006,6231:623108.
[5]姚姚.地震波场与地震勘探[M].北京:地质出版社,2006:22-26.Yao Yao.Seismic Wave Field and Seismic Exploration[M].Beijin:Geological Publishing House,2006:22-26(in Chinese).
[6]贺银芝,沈松,应怀樵,等.小波包分解及其能量谱在发动机连杆轴承故障诊断中的应用[J].振动工程学报,2001,14(1):72-75.He Yinzhi,Shen Song,Ying Huaiqiao,et al.Application of wavelet packet decomposition and its energy spectrum on the fault diagnosis of reciprocation machinery[J].Journal of Vibration Engineering,2001,14(1):72-75(in Chinese).
[7]丁幼亮,李爱群,缪长青.基于小波包能量谱的结构损伤预警方法研究[J].工程力学,2006,23(8):29-38.Ding Youliang,Li Aiqun,Miu Changqing.Investigation on the structural damage alarming method based on wavelet packet energy spectrum[J].Engineering Me-chanics,2006,23(8):29-38(in Chinese).
[8]郑海波,陈心昭,李志远.基于双谱的齿轮故障特征提取与识别[J].振动工程学报,2002,15(4):354-358.Zheng Haibo,Chen Xinzhao,Li Zhiyuan.Bispectrum based gear fault feature extraction and diagnosis[J].Journal of Vibration Engineering,2002,15(4):354-358(in Chinese).
[9]Zhang Guicai,Chen Jin,Li Fucai,et al.Extracting gear fault features using maximal bispectrum[J].Key Engi-neering Materials,2005,293/294:167-174.
[10]曲志刚,靳世久,周琰,等.基于RBF网络的油气管道侵入事件识别方法研究[J].化工自动化及仪表,2007,34(3):58-61.Qu Zhigang,Jin Shijiu,Zhou Yan,et al.Study on recognition method based on RBF neural network for intrusion events of oil and gas pipeline[J].Control and Instruments in Chemical Industry,2007,34(3):58-61(in Chinese).
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心