超声波管道内检测腐蚀缺陷分类识别研究
|
摘要
超声波检测是输油管道在线内检测的重要方法之一,管道腐蚀使管道壁产生大量的突变界面,使超声检测回波信号复杂,其分类是一个高维分类问题。利用支持向量机在解决小样本、非线性、高维模式识别中特有的优势,直接采用表征超声回波形态的A扫描数据作为特征向量,将特征提取与模式分类统一进行,建立了管道腐蚀检测信号分类的二次规划优化模型,并利用优化函数求得该问题的最优解,根据最优解建立了用于回波信号分类判断的分类决策函数,实现管道腐蚀缺陷识别。实验表明基于支持向量机的管道腐蚀超声内检测信号分类识别方法可以分类识别管道腐蚀产生的突变界面和腐蚀裂纹。
The ultrasonic detection is one of the important ways to inspect the wall-loss defects in-line for the oil pipeline. Pipeline corrosion defects make the rough surfaces of pipelines into break interfaces. The interfaces make the ultrasonic echoes complicated.The recognition of the echoes is a high-dimensional classification problem.To solve the problem, a further analysis of ultrasonic echo signals was carried out, showing that the method of measuring pipeline wall thickness with the time interval of the proceeding two ultrasonic echo signals brings about high fall-out and omission ratios.An effective method based on Support Vector Machine (SVM) what is suitable for small-sample,non-linear and high-dimensional recognition for classification and recognition of pipeline corrosion defects was put forward. The ultrasonic A-scan time-series were considered characteristic vectors. With unified considering the characteristic extraction and corrosion defects recognition the quadratic programming problem model of ultrasonic echo signals classification was established.The answer of this problem was got by using the optimization function.The classified decision function of ultrasonic echo signals was established. Experiments show that the classified result of break interfaces of pipelines and corrosive crack is accurate and clear and the method is suitable for in-line detection of pipeline corrosion defects.
The ultrasonic detection is one of the important ways to inspect the wall-loss defects in-line for the oil pipeline. Pipeline corrosion defects make the rough surfaces of pipelines into break interfaces. The interfaces make the ultrasonic echoes complicated.The recognition of the echoes is a high-dimensional classification problem.To solve the problem, a further analysis of ultrasonic echo signals was carried out, showing that the method of measuring pipeline wall thickness with the time interval of the proceeding two ultrasonic echo signals brings about high fall-out and omission ratios.An effective method based on Support Vector Machine (SVM) what is suitable for small-sample,non-linear and high-dimensional recognition for classification and recognition of pipeline corrosion defects was put forward. The ultrasonic A-scan time-series were considered characteristic vectors. With unified considering the characteristic extraction and corrosion defects recognition the quadratic programming problem model of ultrasonic echo signals classification was established.The answer of this problem was got by using the optimization function.The classified decision function of ultrasonic echo signals was established. Experiments show that the classified result of break interfaces of pipelines and corrosive crack is accurate and clear and the method is suitable for in-line detection of pipeline corrosion defects.
引文
[1]李著信,苏毅,吕宏庆,等.管道在线检测技术及检测机器人研究[J].后勤工程学院学报,2006,4:41-45.
[2]戴波,盛沙,唐建,等.改进的Burg最大熵法在管道检测中的应用[J].传感技术学报,2007,6:1416~1419.
[3]V N Vapnik.The nature of statistical learning[M].New York:Springer-verlag,1995.
[4]Nello Cristianini,John Shawe-Taylor.An Instroduction to Support Vector Machines and other Kernel-based Learning Methods[M].北京:电子工业出版社,2004.
[5]王海军,刘贵忠.基于支持向量机的信号自动检测算法[J].地震学报,2007,29(1):85-94.
[6]佘传伏,余立钧,姚道敏.基于特征选择的支持向量机在故障诊断中的应用[J].现代机械,2007,1,22-24.
[7]鹿卫国,戴亚平,高峰.一种基于概率分布估计的水电机组故障预警方法[J].中国电机工程学报,2005,25(4):94-98.
[8]V N Vapnik.Estimation of Dependences Based on Empir-ical Data[M].New York:Springer-Verlag,1982.
[9]汤楠,穆向阳,徐娟.基于超声检测的管道内腐蚀缺陷识别研究[J].石油机械,2005,33(11):54-56.
[10]郑贤斌,陈国明,袁超红.油气管道缺陷检测的数据处理方法回顾与展望[J].压力容器,2005,22(10):38-43.
[11]凌昌荣,刘桂雄,陈铁群.应用集成神经网络识别超声检测缺陷[J].现代制造工程,2006,(4):98-100.
[12]刘旭,夏金东,吴淼.超声检测缺陷分类的降噪及特征提取问题研究[J].中国矿业大学学报:自然科学版,2001,30(3):248-251.
[13]吴德新,杨小林.超声检测中的波形识别与缺陷定性[J].无损检测,2007,24(7):312-316.
[14]Mathias M.Adankon,Mohamed Cheriet.Optimizing re-sources in model selection for support vector machine[J].Pattern Recognition,2007,40:953-963.
[15]Shengfa Yuan,Fulei Chu.Fault diagnosis based on sup-port vector machines with parameter optimization by artifi-cial immunization algorithm[J].Mechanical Systems and Signal Processing,2007,21:1318-1330.
[16]林生梁,刘志.基于RBF核函数的支持向量机参数选择[J].浙江工业大学学报,2007,35(2):163-167.
[17]董春曦,饶鲜,杨绍全,等.支持向量机参数选择方法研究[J].系统工程与电子技术,2004,26(3):1117-1120.
[2]戴波,盛沙,唐建,等.改进的Burg最大熵法在管道检测中的应用[J].传感技术学报,2007,6:1416~1419.
[3]V N Vapnik.The nature of statistical learning[M].New York:Springer-verlag,1995.
[4]Nello Cristianini,John Shawe-Taylor.An Instroduction to Support Vector Machines and other Kernel-based Learning Methods[M].北京:电子工业出版社,2004.
[5]王海军,刘贵忠.基于支持向量机的信号自动检测算法[J].地震学报,2007,29(1):85-94.
[6]佘传伏,余立钧,姚道敏.基于特征选择的支持向量机在故障诊断中的应用[J].现代机械,2007,1,22-24.
[7]鹿卫国,戴亚平,高峰.一种基于概率分布估计的水电机组故障预警方法[J].中国电机工程学报,2005,25(4):94-98.
[8]V N Vapnik.Estimation of Dependences Based on Empir-ical Data[M].New York:Springer-Verlag,1982.
[9]汤楠,穆向阳,徐娟.基于超声检测的管道内腐蚀缺陷识别研究[J].石油机械,2005,33(11):54-56.
[10]郑贤斌,陈国明,袁超红.油气管道缺陷检测的数据处理方法回顾与展望[J].压力容器,2005,22(10):38-43.
[11]凌昌荣,刘桂雄,陈铁群.应用集成神经网络识别超声检测缺陷[J].现代制造工程,2006,(4):98-100.
[12]刘旭,夏金东,吴淼.超声检测缺陷分类的降噪及特征提取问题研究[J].中国矿业大学学报:自然科学版,2001,30(3):248-251.
[13]吴德新,杨小林.超声检测中的波形识别与缺陷定性[J].无损检测,2007,24(7):312-316.
[14]Mathias M.Adankon,Mohamed Cheriet.Optimizing re-sources in model selection for support vector machine[J].Pattern Recognition,2007,40:953-963.
[15]Shengfa Yuan,Fulei Chu.Fault diagnosis based on sup-port vector machines with parameter optimization by artifi-cial immunization algorithm[J].Mechanical Systems and Signal Processing,2007,21:1318-1330.
[16]林生梁,刘志.基于RBF核函数的支持向量机参数选择[J].浙江工业大学学报,2007,35(2):163-167.
[17]董春曦,饶鲜,杨绍全,等.支持向量机参数选择方法研究[J].系统工程与电子技术,2004,26(3):1117-1120.
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