基于频域系统辨识的石油K型井架结构损伤理论研究
|
摘要
石油井架是钻机起升设备的重要组成部分之一,起着安放天车,悬挂游车、大钩及专用工具的重要作用。对于一些在役石油井架由于多年在野外恶劣的环境下长期使用,许多构件会产生不同程度的变形、腐蚀及螺孔扩大,因此,必须对在役石油井架结构进行损伤检测和评估,充分了解石油井架的实际状况。如果能及时发现损伤,并诊断出局部损伤的位置以及损伤程度,就能制定出正确的维修策略,及时修复,恢复结构的承载能力,从而为延长使用寿命,经济可靠地使用现有石油井架提供依据;而且对于避免灾难性事故的发生,保障人们的生命安全财产更为重要。开展石油井架结构损伤识别研究,提高石油井架承载能力检测与安全评价水平,无疑具有重要的理论意义和实用价值。
首先,本文在查阅了大量文献的基础上,综合分析列举了国内外损伤识别技术的研究现状情况,并且介绍了模态实验分析的基本原理以及基于频域系统辨识石油井架结构模态参数的基本理论、基本方法。
其次,本文对基于BP神经网络的损伤识别方法进行了探讨。在此过程中阐述了神经网络的基本概念、特点、模型及其算法,比较详细地研究了BP神经网络、BP经典算法和经过优化的LM算法以及BP神经网络在MATLAB中的实现。
概括的讲,本文针对石油JJ160/41-K型井架的损伤定位与定量问题提出了人工神经网络识别方法,利用有限元软件ANSYS及MATLAB程序将振动模态分析理论和BP神经网络相结合应用于石油井架损伤识别中,同时实现对损伤位置与损伤程度的识别,形成一套基于振动模态分析理论和BP神经网络的石油井架损伤识别方法,取得了较为满意的效果。随着结构模态检测分析仪器精度和参数识别技术的不断提高,基于神经网络的损伤识别方法在工程结构损伤检测中具有很好的应用前景。
Rig derrick is one of the most important equipment in hoist system. People can install crownblock and hang tackle, hoisting hook and special tools on it. Some derricks in using have been used for a long time in the open, many parts of the derrick have distortion and corruption of different extent, and enlarged screw. Thus, we must detect and evaluate the damage and fully realize the actual status of the derrick structure in using. If we can find the damage immediately, diagnose the location and the extent of the damage, constitute the correct servicing strategy, resume the ability of bearing the weight of load, it is good to postpone the natural life of derrick and to provide the proof of using present derrick economically. Furthermore, it is very important to avoid disaster and to protect the people’s lives and estate. The study on damage identification of derrick and making progress in evaluating the ability of load and safety has important realistic value and theoretics significance.
Firstly, on the basis of collection and analysis of much literature, this article analyses the technology of structural damage identification of domestic and foreign research present situation, and introduced the basic theory of modal experiment and the basic theory and method of the derrick’s modal parameter based on frequency domain system identification.
Secondly, the damage identification method based on BP neural networks is discussed and analyzed. All of these such as the basic consept of artificial neural networks, the characteristic, the model, its algorithm are introduced in this process. At the same time, This article elaborates BP neural networks, BP classic algorithm, the Lavender-Marquardt algorithm on the condition of optimization, and BP neural networks that realized in the MATLAB in details.
In general, aiming at the problem on the damage location and extent of JJ160/ 41-K type derrick, the method of damage identification based on artificial neural networks is introduced. Vibration modal analysis theory is integrated with BP neural networks so as to detect derrick’s damage with the help of ANSYS and MATLAB, at the same time, it has been successful in detecting damaged position and extent likely occurring, the method based on vibration modal analysis theory and BP neural networks is satisfied in this paper. Along with the improvement of precision of the modal analytic apparatus and the technology of identifying parameter, the results is that the neural network applied in structure is feasible and have good application perspective.
首先,本文在查阅了大量文献的基础上,综合分析列举了国内外损伤识别技术的研究现状情况,并且介绍了模态实验分析的基本原理以及基于频域系统辨识石油井架结构模态参数的基本理论、基本方法。
其次,本文对基于BP神经网络的损伤识别方法进行了探讨。在此过程中阐述了神经网络的基本概念、特点、模型及其算法,比较详细地研究了BP神经网络、BP经典算法和经过优化的LM算法以及BP神经网络在MATLAB中的实现。
概括的讲,本文针对石油JJ160/41-K型井架的损伤定位与定量问题提出了人工神经网络识别方法,利用有限元软件ANSYS及MATLAB程序将振动模态分析理论和BP神经网络相结合应用于石油井架损伤识别中,同时实现对损伤位置与损伤程度的识别,形成一套基于振动模态分析理论和BP神经网络的石油井架损伤识别方法,取得了较为满意的效果。随着结构模态检测分析仪器精度和参数识别技术的不断提高,基于神经网络的损伤识别方法在工程结构损伤检测中具有很好的应用前景。
Rig derrick is one of the most important equipment in hoist system. People can install crownblock and hang tackle, hoisting hook and special tools on it. Some derricks in using have been used for a long time in the open, many parts of the derrick have distortion and corruption of different extent, and enlarged screw. Thus, we must detect and evaluate the damage and fully realize the actual status of the derrick structure in using. If we can find the damage immediately, diagnose the location and the extent of the damage, constitute the correct servicing strategy, resume the ability of bearing the weight of load, it is good to postpone the natural life of derrick and to provide the proof of using present derrick economically. Furthermore, it is very important to avoid disaster and to protect the people’s lives and estate. The study on damage identification of derrick and making progress in evaluating the ability of load and safety has important realistic value and theoretics significance.
Firstly, on the basis of collection and analysis of much literature, this article analyses the technology of structural damage identification of domestic and foreign research present situation, and introduced the basic theory of modal experiment and the basic theory and method of the derrick’s modal parameter based on frequency domain system identification.
Secondly, the damage identification method based on BP neural networks is discussed and analyzed. All of these such as the basic consept of artificial neural networks, the characteristic, the model, its algorithm are introduced in this process. At the same time, This article elaborates BP neural networks, BP classic algorithm, the Lavender-Marquardt algorithm on the condition of optimization, and BP neural networks that realized in the MATLAB in details.
In general, aiming at the problem on the damage location and extent of JJ160/ 41-K type derrick, the method of damage identification based on artificial neural networks is introduced. Vibration modal analysis theory is integrated with BP neural networks so as to detect derrick’s damage with the help of ANSYS and MATLAB, at the same time, it has been successful in detecting damaged position and extent likely occurring, the method based on vibration modal analysis theory and BP neural networks is satisfied in this paper. Along with the improvement of precision of the modal analytic apparatus and the technology of identifying parameter, the results is that the neural network applied in structure is feasible and have good application perspective.
引文
[1] 姜绍飞.基于神经网络的结构优化与损伤探测[M].北京:科学出版社,2002.3~4.
[2] 孙宗光.大跨度斜拉桥结构的动力损伤检测[D].浙江大学, 2001, 6:3～9.
[3] Scott W Doebling.A Summary Review of Vibration-Based Damage Identification Methods.The Shock and Vibration Digest[J], 1998(3):91~105.
[4] Popovics J Setal.A survey of developments in ultrasonic NDE of concrete[J], IEEE Transactions on Ultrasonic Ferroelectrics and Frequency Control[J], 1994,41(1):140~143.
[5] Zalameda J Netal.A field deployable nondestructive impact damage assessment methodology for composite structures[J], Journal of Composite Technology&Research 1994,16(2):161～169.
[6] Jenks W G et al.Squids for nondestructive evaluation[J], Journal of Physics&Applied Physics,1997,30(3):293～323.
[7] Vandiver j K.Detection of structural failure in fixed platforms by measurement of dynamic response[J], J Petroleum Tech.,1977,29:305～310.
[8] 刘利军.基于模态分析理论的桥梁结构损伤识别研究[D].武汉理工大学硕士学位论文.
[9] 郑栋梁,李中付.结构早期损伤识别技术的现状和发展趋势[J].振动与冲击,2002, 21(2):1～6.
[10] Cawley P, Adams R D.The Iocation of Defects in structures from Measurements of Natural Frequencies[J].Journal of strain anlysis,1979,14(2):7～12.
[11] Cawley P, Adams R D.A Vibration Technique for Non-Destructive Testing of Fibre Composite Structures[J].Journal of Composite Materials, April 1979,13(3):9～14.
[12] Adams R D, Cawley P.A Vibration Technique for Non-Destructive Assessing the Integrity of Structures[J].Journal of Mechanicals Engineering Science ,1978,20(2):148～152.
[13] 高文清,金建明,高淑英.基于模态分析的结构损伤检测方法研究[J].西南交通大学学报,1998, 33(1):108～113.
[14] 郭国会,易伟建.基于频率进行简支梁损伤评估的数值方法[J].重庆建筑大学学报,2001,23(2):17～21.
[15] Fabricio Vesvroni,Davila Cape chi.Damage Detection in structures Based on Frequently Measurements[J].Journal of Engineering Mechanics, July 2000.
[16] Salawu, OS.Detection of structure damage through changes in frequency[J] . a review.Engineering structures ,1997, 19(2):718～723.
[17] 邹晓军.梁桥结构损伤识别的曲率模态技术[D].武汉理工大学,2003,4:20～49.
[18] 董聪.基于动力特性的结构损伤定位方法[J].力学与实践,1999,21(4):62～63.
[19] 李德葆.实验应变/应力模态分析若干问题的进展评述[J].振动与冲击, 1996, 30(l):13～17.
[20] 张开银.板结构裂纹故障的模式识别[J].武汉汽车工业大学学报,1997,20(3).82～86.
[21] 董聪,丁辉,高嵩.结构损伤识别和定位的基本原理与方法[J].中国铁道科学, 1999,20(3):89～94.
[22] 董聪.结构系统故障诊断的理论与方法[J].强度与环境,2000(1):1～16.
[23] 任权.基于应变模态变化率的压力管道无损检测[J].大连理工大学学报,2001, 4l(6):648～652.
[24] 唐小兵.结构裂纹位置识别的模态分析[J].武汉理工大学学报,2001,23(5):71～74.
[25] 史治宇,吕令毅.由模态应变能法诊断结构破损的实验研究[J].东南大学学报,1999, 29(2):134～138.
[26] 史治宇,张令弥,吕令毅.基于模态应变能诊断结构破损的修正方法[J].东南大学学报,2000, 30(3):84～87.
[27] 伊娟.基于柔度改变的连续梁桥损伤诊断研究[J].华东公路,2001(3):25～27.
[28] 蔡贤辉.一种桁架结构损伤识别的柔度阵法[J].计算力学学报,2001,18(1):42～47.
[29] 韩西.用矩阵结构模型进行结构损伤检测和故障诊断[J].重庆交通学院学报,2002,21(2):108～110.
[30] 秦权.悬索桥的损伤识别[J].清华大学学报,1998,38(12):44～47.
[31] 王代华.基于神经网络的柔性结构损伤模式识别方法[J].振动、测试与诊断,2000,20(4):240～244.
[32] 陆秋海.利用模态试验参数识别结构损伤的神经网络法[J]. 工程力学, 1999,16(1):35～42 .
[33] 苏娟.神经网络法在定量损伤识别研究中的应用[J].清华大学报,1999, 39(4):68～70.
[34] 邱洪兴,蒋永生.结构损伤区域的判断分析法[J].工业建筑,2000, 30(4):61～63.
[35] 张蓓,殷学纲.结构损伤的概率诊断[J].重庆大学学报.2000, 23,(2):60～63.
[36] 张开鹏.结构损伤识别方法研究[D].武汉理工大学,2004.
[37] S·K·Thyagarajan et al . Detecting Structural Damage Using Frequency Response Functions. Journal of Sound and Vibration[J], 1998,210(1):162～170.
[38] Mark J S etal.Detecting Structural Damage Using Transmittance Function. Proceeding of 15th IMAC ,Florida[J],1997,32(2):638～644.
[39] 郑明刚.基于频响函数的结构损伤检测[J].机械科学与技术,1999,16(1):35～42.
[40] 李晏,石来德.利用应变传递特性诊断箱型梁裂纹的实验研究[J].同济大学学报,1999,27(5):618～620.
[41] 易伟建.基于遗传算法的结构损伤诊断研究[J].工程力学,2001,18(2):64～71.
[42] 高宝成.基于小波分析的简支梁裂缝识别方法研究[J].Journal of Vibration Engineering,1997,10(3):81～85.
[43] 王中要.用残余力向量进行连续梁损伤诊断[J].昆明理工大学学报,2000, 25(5):64～67.
[44] 李守巨.基于混合优化策略的结构损伤识别方法[J].中国矿业大学学报,2002, 31(1):39～43.
[45] 高宝成.一种混凝土简支梁裂缝识别方法研究[J].华中理工大学学报,1997,25(增刊1):15～17.
[46] 张启伟,范立础.利用动静力测量数据的桥梁结构损伤识别[J].同济大学学报,1998, 26(5):528～532.
[47] 张启伟.桥梁结构模型修正技术与损伤识别[D].上海:同济大学博士学位论文,1999.
[48] 梁冰.网架损伤识别的动测研究[D].武汉理工大学,2005.
[49] 袁万城,崔飞,张启伟.桥梁健康检测与状态评估的研究现状与发展[J].同济大学学报,1999,27(2),184～188.
[50] 周传荣,赵淳生.机械振动参数识别及其应用[M].北京:科学出版社,1989.60～66.
[51] 蒋济同.海洋平台的模态参数识别与损伤诊断研究[D] .中国海洋大学,2003,5:66～70.
[52] 懂长虹.Matlab 神经网络与应用[M].国防工业出版社,2005.106～178.
[53] 丛爽.面向 MAYLAB 工具箱的神经网络理论与应用[M].合肥:中国科学技术大学出版社,1998.62～72.
[54] 王柏生.框架结构连接损伤识别神经网络输入参数的确定[J].振动工程学报, 2000,13(1):137～142.
[55] Cawley P, Adams R D. T he location of defects in structures from measurements of natural frequencies.Journal of Strain Analysis[J] ,1979,14(2):49～57.
[56] Lam H F, Ko J M,Wong C W.Localization of damaged structural connections based on experimental modal and sensitivity analysis.Journal of Sound and Vibration[J] ,1998,210(1):91～115.
[57] 焦李成.神经网络系统理论[M].西安:西安电子科技大学出版社,1990.78～88.
[58] 焦李成.神经网络的应用与实现[M].西安:西安电子科技大学出版社,1993.69～72. [59」张立明.人工神经网络的模型及其应用[M].复旦大学出版社,1992.55～76.
[60] 洛伦茨 GG 著.谢庭藩,施威亮译.函数通近论[M].上海:上海科学技术出版社,1981.34～42.
[61] 邹龙庆.JJ160/41-K 型井架有限元分析与承载能力研究[J] .石油矿场机械,2004, 33(6):33～35.
[2] 孙宗光.大跨度斜拉桥结构的动力损伤检测[D].浙江大学, 2001, 6:3～9.
[3] Scott W Doebling.A Summary Review of Vibration-Based Damage Identification Methods.The Shock and Vibration Digest[J], 1998(3):91~105.
[4] Popovics J Setal.A survey of developments in ultrasonic NDE of concrete[J], IEEE Transactions on Ultrasonic Ferroelectrics and Frequency Control[J], 1994,41(1):140~143.
[5] Zalameda J Netal.A field deployable nondestructive impact damage assessment methodology for composite structures[J], Journal of Composite Technology&Research 1994,16(2):161～169.
[6] Jenks W G et al.Squids for nondestructive evaluation[J], Journal of Physics&Applied Physics,1997,30(3):293～323.
[7] Vandiver j K.Detection of structural failure in fixed platforms by measurement of dynamic response[J], J Petroleum Tech.,1977,29:305～310.
[8] 刘利军.基于模态分析理论的桥梁结构损伤识别研究[D].武汉理工大学硕士学位论文.
[9] 郑栋梁,李中付.结构早期损伤识别技术的现状和发展趋势[J].振动与冲击,2002, 21(2):1～6.
[10] Cawley P, Adams R D.The Iocation of Defects in structures from Measurements of Natural Frequencies[J].Journal of strain anlysis,1979,14(2):7～12.
[11] Cawley P, Adams R D.A Vibration Technique for Non-Destructive Testing of Fibre Composite Structures[J].Journal of Composite Materials, April 1979,13(3):9～14.
[12] Adams R D, Cawley P.A Vibration Technique for Non-Destructive Assessing the Integrity of Structures[J].Journal of Mechanicals Engineering Science ,1978,20(2):148～152.
[13] 高文清,金建明,高淑英.基于模态分析的结构损伤检测方法研究[J].西南交通大学学报,1998, 33(1):108～113.
[14] 郭国会,易伟建.基于频率进行简支梁损伤评估的数值方法[J].重庆建筑大学学报,2001,23(2):17～21.
[15] Fabricio Vesvroni,Davila Cape chi.Damage Detection in structures Based on Frequently Measurements[J].Journal of Engineering Mechanics, July 2000.
[16] Salawu, OS.Detection of structure damage through changes in frequency[J] . a review.Engineering structures ,1997, 19(2):718～723.
[17] 邹晓军.梁桥结构损伤识别的曲率模态技术[D].武汉理工大学,2003,4:20～49.
[18] 董聪.基于动力特性的结构损伤定位方法[J].力学与实践,1999,21(4):62～63.
[19] 李德葆.实验应变/应力模态分析若干问题的进展评述[J].振动与冲击, 1996, 30(l):13～17.
[20] 张开银.板结构裂纹故障的模式识别[J].武汉汽车工业大学学报,1997,20(3).82～86.
[21] 董聪,丁辉,高嵩.结构损伤识别和定位的基本原理与方法[J].中国铁道科学, 1999,20(3):89～94.
[22] 董聪.结构系统故障诊断的理论与方法[J].强度与环境,2000(1):1～16.
[23] 任权.基于应变模态变化率的压力管道无损检测[J].大连理工大学学报,2001, 4l(6):648～652.
[24] 唐小兵.结构裂纹位置识别的模态分析[J].武汉理工大学学报,2001,23(5):71～74.
[25] 史治宇,吕令毅.由模态应变能法诊断结构破损的实验研究[J].东南大学学报,1999, 29(2):134～138.
[26] 史治宇,张令弥,吕令毅.基于模态应变能诊断结构破损的修正方法[J].东南大学学报,2000, 30(3):84～87.
[27] 伊娟.基于柔度改变的连续梁桥损伤诊断研究[J].华东公路,2001(3):25～27.
[28] 蔡贤辉.一种桁架结构损伤识别的柔度阵法[J].计算力学学报,2001,18(1):42～47.
[29] 韩西.用矩阵结构模型进行结构损伤检测和故障诊断[J].重庆交通学院学报,2002,21(2):108～110.
[30] 秦权.悬索桥的损伤识别[J].清华大学学报,1998,38(12):44～47.
[31] 王代华.基于神经网络的柔性结构损伤模式识别方法[J].振动、测试与诊断,2000,20(4):240～244.
[32] 陆秋海.利用模态试验参数识别结构损伤的神经网络法[J]. 工程力学, 1999,16(1):35～42 .
[33] 苏娟.神经网络法在定量损伤识别研究中的应用[J].清华大学报,1999, 39(4):68～70.
[34] 邱洪兴,蒋永生.结构损伤区域的判断分析法[J].工业建筑,2000, 30(4):61～63.
[35] 张蓓,殷学纲.结构损伤的概率诊断[J].重庆大学学报.2000, 23,(2):60～63.
[36] 张开鹏.结构损伤识别方法研究[D].武汉理工大学,2004.
[37] S·K·Thyagarajan et al . Detecting Structural Damage Using Frequency Response Functions. Journal of Sound and Vibration[J], 1998,210(1):162～170.
[38] Mark J S etal.Detecting Structural Damage Using Transmittance Function. Proceeding of 15th IMAC ,Florida[J],1997,32(2):638～644.
[39] 郑明刚.基于频响函数的结构损伤检测[J].机械科学与技术,1999,16(1):35～42.
[40] 李晏,石来德.利用应变传递特性诊断箱型梁裂纹的实验研究[J].同济大学学报,1999,27(5):618～620.
[41] 易伟建.基于遗传算法的结构损伤诊断研究[J].工程力学,2001,18(2):64～71.
[42] 高宝成.基于小波分析的简支梁裂缝识别方法研究[J].Journal of Vibration Engineering,1997,10(3):81～85.
[43] 王中要.用残余力向量进行连续梁损伤诊断[J].昆明理工大学学报,2000, 25(5):64～67.
[44] 李守巨.基于混合优化策略的结构损伤识别方法[J].中国矿业大学学报,2002, 31(1):39～43.
[45] 高宝成.一种混凝土简支梁裂缝识别方法研究[J].华中理工大学学报,1997,25(增刊1):15～17.
[46] 张启伟,范立础.利用动静力测量数据的桥梁结构损伤识别[J].同济大学学报,1998, 26(5):528～532.
[47] 张启伟.桥梁结构模型修正技术与损伤识别[D].上海:同济大学博士学位论文,1999.
[48] 梁冰.网架损伤识别的动测研究[D].武汉理工大学,2005.
[49] 袁万城,崔飞,张启伟.桥梁健康检测与状态评估的研究现状与发展[J].同济大学学报,1999,27(2),184～188.
[50] 周传荣,赵淳生.机械振动参数识别及其应用[M].北京:科学出版社,1989.60～66.
[51] 蒋济同.海洋平台的模态参数识别与损伤诊断研究[D] .中国海洋大学,2003,5:66～70.
[52] 懂长虹.Matlab 神经网络与应用[M].国防工业出版社,2005.106～178.
[53] 丛爽.面向 MAYLAB 工具箱的神经网络理论与应用[M].合肥:中国科学技术大学出版社,1998.62～72.
[54] 王柏生.框架结构连接损伤识别神经网络输入参数的确定[J].振动工程学报, 2000,13(1):137～142.
[55] Cawley P, Adams R D. T he location of defects in structures from measurements of natural frequencies.Journal of Strain Analysis[J] ,1979,14(2):49～57.
[56] Lam H F, Ko J M,Wong C W.Localization of damaged structural connections based on experimental modal and sensitivity analysis.Journal of Sound and Vibration[J] ,1998,210(1):91～115.
[57] 焦李成.神经网络系统理论[M].西安:西安电子科技大学出版社,1990.78～88.
[58] 焦李成.神经网络的应用与实现[M].西安:西安电子科技大学出版社,1993.69～72. [59」张立明.人工神经网络的模型及其应用[M].复旦大学出版社,1992.55～76.
[60] 洛伦茨 GG 著.谢庭藩,施威亮译.函数通近论[M].上海:上海科学技术出版社,1981.34～42.
[61] 邹龙庆.JJ160/41-K 型井架有限元分析与承载能力研究[J] .石油矿场机械,2004, 33(6):33～35.