基于LabVIEW的汽轮机在线监测与故障诊断系统
|
摘要
汽轮发电机组属于典型的旋转机械,旋转机械振动信号从幅值域、频率域和时间域实时地反映了机器故障信息。因此,了解和掌握汽轮机在故障状态下的振动特征,在监测机器的运行状态和提高诊断故障的准确度方面具有重要的理论意义和实际工程应用价值。
根据机械振动理论和旋转机械产生的故障机理,可以知道绝大多数旋转机械的故障征兆都有其相对应的振动特征,通过对振动信号的分析处理可以将这些振动特征提取出来。采用傅立叶变换对非平稳振动信号进行频率谱分析,频谱分析的结果只是在整个被分析时间段上的平均,不能反映信号突变的细节,也就无法对转子故障诊断做出准确判断。同时,因早期故障冲击幅值较小,易被周围环境其他振动干扰,导致信号信噪比较低。所以,对信号进行信噪分离、提取故障信息是转子故障诊断的关键。
小波分析是近年来发展非常迅猛的时频分析方法。由于其对信号去噪、还原性都较好,特别适用于对含有大量背景噪音的信噪比非常低的信号分析和调理。本论文利用虚拟仪器软件实现小波分析,将监测振动信号在全监测频带范围内逐层分解为若干频带,采用Matlab构建BP神经网络,提取训练好的神经网络的权阈,将转子发生故障部件的特征频率作为神经网络的输入,经过神经网络的并行数值计算,输出对应的故障,在LabVIEW下编程实现了基于神经网络的故障诊断程序,对所设计的系统进行测试,并对模拟的振动信号进行诊断,实现从故障征兆空间到故障空间的映射,表明该系统能够较好地反映设备的运行状态,且具有稳定可靠、实时性好等优越性,具有较高的实用价值。
A turbine generator is a typical circumvolve machine. The vibration signals of a circumvolve machine reflect the fault information in the amplitude domain, the frequency domain, and the time domain. Therefore, understanding and mastering the vibration features of the turbine generator in the faulty status have an important theoretical significance and a practical engineering application value in monitoring the running status of the machine and heightening the accuracy of diagnosing faults.
The mechanical vibration theory and fault mechanism of the circumvolve machine shows that most fault symptoms have the corresponding vibration features. Through the analysis and disposal of the vibration signals, the vibration features can be drawn out. The spectrum analysis on the non-stationary vibration signals by using the Fourier transform shows that the signals keep stationary on the analyzed period but the specific wavelets transform cannot be reflected, thus leading to inaccurate identification of the fault in the rotor. The early fault impact amplitude value is small, so it is easily interfered by other vibrations, thus leading to a low signal-to-noise ratio. In this sense, the key to diagnosing the fault in the rotor is to separate signals and noises and withdraw the faulty information.
Wavelet analysis is a very popular method for time frequency analysis. It can be used to eliminate the noise from the signal and restore the signal, so it is applicable for analyzing and suiting the signals with a very low signal-to-noise ratio. This dissertation uses virtual apparatus software to analyze the wavelets. The monitored vibration signals are first decomposed into several frequency bands within the full-monitoring frequency band. We design a BP network of nerve by Matlab and distill the coefficient and threshold of the network, which were programmed by the LabVIEW to analysis the vibration signals. The characteristic frequency of the faulty components of the rotor is used as the input of the nerve network, and then through calculating the parallel values of the nerve network the corresponding fault is output to realize the mapping between the fault symptom space and fault space, thus realizing the identification and diagnosis of faults.
We make the fault signal generator to measure the system that we design, it indicates that system can reflect the state of the machine. It operates steadily reliably and practically. It has a lot of practical worth.
根据机械振动理论和旋转机械产生的故障机理,可以知道绝大多数旋转机械的故障征兆都有其相对应的振动特征,通过对振动信号的分析处理可以将这些振动特征提取出来。采用傅立叶变换对非平稳振动信号进行频率谱分析,频谱分析的结果只是在整个被分析时间段上的平均,不能反映信号突变的细节,也就无法对转子故障诊断做出准确判断。同时,因早期故障冲击幅值较小,易被周围环境其他振动干扰,导致信号信噪比较低。所以,对信号进行信噪分离、提取故障信息是转子故障诊断的关键。
小波分析是近年来发展非常迅猛的时频分析方法。由于其对信号去噪、还原性都较好,特别适用于对含有大量背景噪音的信噪比非常低的信号分析和调理。本论文利用虚拟仪器软件实现小波分析,将监测振动信号在全监测频带范围内逐层分解为若干频带,采用Matlab构建BP神经网络,提取训练好的神经网络的权阈,将转子发生故障部件的特征频率作为神经网络的输入,经过神经网络的并行数值计算,输出对应的故障,在LabVIEW下编程实现了基于神经网络的故障诊断程序,对所设计的系统进行测试,并对模拟的振动信号进行诊断,实现从故障征兆空间到故障空间的映射,表明该系统能够较好地反映设备的运行状态,且具有稳定可靠、实时性好等优越性,具有较高的实用价值。
A turbine generator is a typical circumvolve machine. The vibration signals of a circumvolve machine reflect the fault information in the amplitude domain, the frequency domain, and the time domain. Therefore, understanding and mastering the vibration features of the turbine generator in the faulty status have an important theoretical significance and a practical engineering application value in monitoring the running status of the machine and heightening the accuracy of diagnosing faults.
The mechanical vibration theory and fault mechanism of the circumvolve machine shows that most fault symptoms have the corresponding vibration features. Through the analysis and disposal of the vibration signals, the vibration features can be drawn out. The spectrum analysis on the non-stationary vibration signals by using the Fourier transform shows that the signals keep stationary on the analyzed period but the specific wavelets transform cannot be reflected, thus leading to inaccurate identification of the fault in the rotor. The early fault impact amplitude value is small, so it is easily interfered by other vibrations, thus leading to a low signal-to-noise ratio. In this sense, the key to diagnosing the fault in the rotor is to separate signals and noises and withdraw the faulty information.
Wavelet analysis is a very popular method for time frequency analysis. It can be used to eliminate the noise from the signal and restore the signal, so it is applicable for analyzing and suiting the signals with a very low signal-to-noise ratio. This dissertation uses virtual apparatus software to analyze the wavelets. The monitored vibration signals are first decomposed into several frequency bands within the full-monitoring frequency band. We design a BP network of nerve by Matlab and distill the coefficient and threshold of the network, which were programmed by the LabVIEW to analysis the vibration signals. The characteristic frequency of the faulty components of the rotor is used as the input of the nerve network, and then through calculating the parallel values of the nerve network the corresponding fault is output to realize the mapping between the fault symptom space and fault space, thus realizing the identification and diagnosis of faults.
We make the fault signal generator to measure the system that we design, it indicates that system can reflect the state of the machine. It operates steadily reliably and practically. It has a lot of practical worth.
引文
[1]施维新.汽轮发电机组振动[M].水利电力出版社,2001.
[2]李录平.汽轮机组故诊断技术[M].中国电力出版社,2001.
[3]赵光宙,舒勤.信号分析与处理[M]. 2001,8
[4]虞和济,陈长征,张省等.基于神经网络的智能诊断[M].北京:冶金工业出版社,2000.5.
[5]第二届全国机械设备故障诊断学术会议论文集[C] . 1988.6.7-10
[6] Muszynska,A. Vibrational Diagnostics of Rotating Machinery Malfunctions. Vki, Vibration and Rotor Dynamics.[C].1992.p40
[7]周志华,曹存根.神经网络及其应用[M],清华大学出版社,2004
[8]刘君华.基于LabVIEW的虚拟仪器设计[M].电子工业出版社,2003.
[9]陈进机械设备振动监测与故障诊断[M].上海交通大学出版社,1999.4.
[10]华中电网有限公司培训中心300MW火电机组集控运行[M].中国电力出版社
[11]张贤达.现代信号处理[M].清华大学出版社,1997
[12]张国华,张文娟,薛鹏翔.小波分析与应用基础[M].西北工业大学出版社,2006
[13] Javier R.Fonollosa,Chrysostomos L,Nikias.,Wiger hider order moment spectra: definition, properties, Computation and application to transient signal analysis[J]. IEEE Trans on signal process,Vol.41(1).
[14] S.Mallat and W.L.Hwang. Singularity detection and processing with wavelet. IEEET ransaction on Information Theory.1992.38(2):617-643
[15] Mallat S.A theory for multiresolution signal decomposition:the wavelet representation.IEEE Trans[J]. On Pattern Analysis and Machine Intelligence, 1989,11(7):674-693.
[16] S.G.Mallat,W.L.Hwang, Singularity detection and proceeding with wavelet[J], IEEE Trans on Information Theory,1992,38(2).
[17]宋友,转子系统碰摩故障诊断及振动信号的降噪研究[D],北京航空航天大学博士学位论文,2003.
[18] Jun Zhang, Ruixin Li, Pu Han, Dongfeng Wang, Xichao Yin, Wavelet Packet Feature Extraction for Vibration Monitoring and Fault Diagnostic of Turbo-Generator.The Second International Conference on Machine Learningand Cybernetics(ICMLC2003),Volume:1,2-5 Nov.2003 Pages:76-81 Vol.1
[19] B Liu.Selection of wavelet packet basis for rotating machinery fault diagnosis[J]. Journal of Sound and Vibration.2005,7.284(3):567-582
[20]裴俊峰杨其俊机械设备股故障诊断技术[M],石油大学出版社, 1997.4
[21] Chris K, Mechefske. Automatic Machinery Fault Detection and Diagnosis Using Fuzzy Logic[C]. Condition Monitoring 96 Conference, USA, 1996.
[22] Chris K, Mechefske.Automatic Machinery Fault Detection and Diagnosis Using Fuzzy Logic[C]. Condition Monitoring 96 Conference, USA, 1996.
[23]汪健,倪维斗人工神经网络用于电厂故障诊断[J].动力工程1997第17卷第2期uerts、81(2003)2165~2172
[24]袁曾任编著.人工神经网络及其应用[M].广西科学技术出版社,1999.10
[25] SchiffT,Theequipment Reliability Program at Willamette Industries Kingsport, TNPlant[J]. Willamette Industries,1998
[26] Hayashi S,Askauar T,Sheng Zhang、Sutdy of maehine afult diagnosis system using neural Networks based on response-only data[J]. Computer and Structures 81(2003)2165-2172
[27]胡守仁等.神经网络导论[M].北京:国防科技大学出版社,1992
[28] Li Yong (李勇)Ye Rongxue(叶荣学),Cao Zuqing(曹祖庆). Improved BP Algorithm Method And Its Application In failure of Turbogenerator[J]. Set.Proceedings of ICTD'95,1995
[29]崔宝珍,潘宏侠.小波分析在滚动轴承敌障诊断中的应用[J].科技情报开发与经济.2005,15(2):176～177.
[30] M.Karpenko,N.Sepehri,D.Scuse Diagnosis of process valve actuator faults using a multilayer neural network[J]. Control Engineering Practice 11(2003)1289-1299.
[31] Adam Adgar, Chris Cox, John MacIntyre. Automatic Fault Diagnosis for Machinery using Statistical and Neural Net-work Techniques[J]. COMADEM Proceedings,1998
[32]周爱红.汽轮机状态监测与故障智能诊断系统[D].重庆大学硕士学位论文. 2003.
[33]张瑞林.机械的状态监测和故障诊断技术[M].化工机械手册北京:化工工业出版社, 1989.
[34]孙传友,孙晓斌.感测技术基础[M].电子工业大学出版社.2001.
[35] Symposium,Redl,Will.Noise and vibration measurement[M].The Society.1985.
[36]张智焕.计算机检测与控制[M].浙江科学技术出版社.2004,6.
[37]孙晓云,郭立炜,孙会琴.基于LabWindows/CVI的虚拟仪器设计与应用[M].电子工业出版社.2005,7.
[2]李录平.汽轮机组故诊断技术[M].中国电力出版社,2001.
[3]赵光宙,舒勤.信号分析与处理[M]. 2001,8
[4]虞和济,陈长征,张省等.基于神经网络的智能诊断[M].北京:冶金工业出版社,2000.5.
[5]第二届全国机械设备故障诊断学术会议论文集[C] . 1988.6.7-10
[6] Muszynska,A. Vibrational Diagnostics of Rotating Machinery Malfunctions. Vki, Vibration and Rotor Dynamics.[C].1992.p40
[7]周志华,曹存根.神经网络及其应用[M],清华大学出版社,2004
[8]刘君华.基于LabVIEW的虚拟仪器设计[M].电子工业出版社,2003.
[9]陈进机械设备振动监测与故障诊断[M].上海交通大学出版社,1999.4.
[10]华中电网有限公司培训中心300MW火电机组集控运行[M].中国电力出版社
[11]张贤达.现代信号处理[M].清华大学出版社,1997
[12]张国华,张文娟,薛鹏翔.小波分析与应用基础[M].西北工业大学出版社,2006
[13] Javier R.Fonollosa,Chrysostomos L,Nikias.,Wiger hider order moment spectra: definition, properties, Computation and application to transient signal analysis[J]. IEEE Trans on signal process,Vol.41(1).
[14] S.Mallat and W.L.Hwang. Singularity detection and processing with wavelet. IEEET ransaction on Information Theory.1992.38(2):617-643
[15] Mallat S.A theory for multiresolution signal decomposition:the wavelet representation.IEEE Trans[J]. On Pattern Analysis and Machine Intelligence, 1989,11(7):674-693.
[16] S.G.Mallat,W.L.Hwang, Singularity detection and proceeding with wavelet[J], IEEE Trans on Information Theory,1992,38(2).
[17]宋友,转子系统碰摩故障诊断及振动信号的降噪研究[D],北京航空航天大学博士学位论文,2003.
[18] Jun Zhang, Ruixin Li, Pu Han, Dongfeng Wang, Xichao Yin, Wavelet Packet Feature Extraction for Vibration Monitoring and Fault Diagnostic of Turbo-Generator.The Second International Conference on Machine Learningand Cybernetics(ICMLC2003),Volume:1,2-5 Nov.2003 Pages:76-81 Vol.1
[19] B Liu.Selection of wavelet packet basis for rotating machinery fault diagnosis[J]. Journal of Sound and Vibration.2005,7.284(3):567-582
[20]裴俊峰杨其俊机械设备股故障诊断技术[M],石油大学出版社, 1997.4
[21] Chris K, Mechefske. Automatic Machinery Fault Detection and Diagnosis Using Fuzzy Logic[C]. Condition Monitoring 96 Conference, USA, 1996.
[22] Chris K, Mechefske.Automatic Machinery Fault Detection and Diagnosis Using Fuzzy Logic[C]. Condition Monitoring 96 Conference, USA, 1996.
[23]汪健,倪维斗人工神经网络用于电厂故障诊断[J].动力工程1997第17卷第2期uerts、81(2003)2165~2172
[24]袁曾任编著.人工神经网络及其应用[M].广西科学技术出版社,1999.10
[25] SchiffT,Theequipment Reliability Program at Willamette Industries Kingsport, TNPlant[J]. Willamette Industries,1998
[26] Hayashi S,Askauar T,Sheng Zhang、Sutdy of maehine afult diagnosis system using neural Networks based on response-only data[J]. Computer and Structures 81(2003)2165-2172
[27]胡守仁等.神经网络导论[M].北京:国防科技大学出版社,1992
[28] Li Yong (李勇)Ye Rongxue(叶荣学),Cao Zuqing(曹祖庆). Improved BP Algorithm Method And Its Application In failure of Turbogenerator[J]. Set.Proceedings of ICTD'95,1995
[29]崔宝珍,潘宏侠.小波分析在滚动轴承敌障诊断中的应用[J].科技情报开发与经济.2005,15(2):176～177.
[30] M.Karpenko,N.Sepehri,D.Scuse Diagnosis of process valve actuator faults using a multilayer neural network[J]. Control Engineering Practice 11(2003)1289-1299.
[31] Adam Adgar, Chris Cox, John MacIntyre. Automatic Fault Diagnosis for Machinery using Statistical and Neural Net-work Techniques[J]. COMADEM Proceedings,1998
[32]周爱红.汽轮机状态监测与故障智能诊断系统[D].重庆大学硕士学位论文. 2003.
[33]张瑞林.机械的状态监测和故障诊断技术[M].化工机械手册北京:化工工业出版社, 1989.
[34]孙传友,孙晓斌.感测技术基础[M].电子工业大学出版社.2001.
[35] Symposium,Redl,Will.Noise and vibration measurement[M].The Society.1985.
[36]张智焕.计算机检测与控制[M].浙江科学技术出版社.2004,6.
[37]孙晓云,郭立炜,孙会琴.基于LabWindows/CVI的虚拟仪器设计与应用[M].电子工业出版社.2005,7.