基于小波包变换和Elman人工神经网络的电机故障诊断系统的研究
|
摘要
随着现代工业的飞速发展,尤其是流水线技术在工业生产中被广泛应用之后,电机已经成为了现代工业技术发展的重要基础。而对于现代电机的设计不仅仅是如何提高其驱动能力的问题,同时其工作的安全性、稳定性和可靠性也成为电机运行过程中不可忽视的重要层面。因此如何对电机的工作状况尤其是工作过程中发生的故障进行有效的模式识别将对工业生产过程稳定有序的进行造成重要影响。
本文在总结了传统的电机故障诊断方法的基础上,通过对电机工作中振动信号的采集与监测以及对电机工作故障的分析,设计了一种基于小波包变换与Elman人工神经网络的电机故障诊断系统,通过小波包变换对采集数据进行信号处理与特征值提取,并利用神经网络的模式识别能力对电机的工作状况进行判定。
本文研究分析了电机在工作过程中常见的工作状况,并针对外壳破裂、基座松脱、转子不对中等三种常见工作故障模式以及正常工作状况通过传感器采集两组不同的振动信号。一组用于对神经网络进行训练,作为样本信号；另一组用于对训练好的神经网络进行性能测试,作为测试信号。对于用于训练学习的振动信号用小波包变换的方法对信号进行特征值提取得到信号的特征向量,并对神经网络系统进行训练。对测试信号进行同样的特征向量提取,并通过训练好的神经网络对电机的工作状况进行诊断。
本文对上述所设计的诊断系统在Matlab平台上进行了系统仿真,验证了算法的有效性和准确性。测试结果符合实际测试信号对应的不同状态,结果证明了本文中所设计的基于小波包变换和Elman人工神经网络电机故障诊断系统可以有效的对直流电动机在工作过程中发生的故障做出有效的诊断。
最后本文对设计的诊断系统进行了必要的总结,并对课题未来延伸的研究方向进行了展望。
With the rapid development of modern industry, particularly after production line is widely used in industrial, the motor has become an important base of modern industrial technology. Modern motor design has not only been driven by the way how to improve their capacity, while the security, stability and reliability have also become the significant aspect cannot be ignored. Therefore, how to effectively recognize the failure occurred during the industrial production will be an important influence for the stable and orderly conduct of the process.
In this work, based on the summary of traditional motor fault diagnosis methods and the analysis of acquisition and monitoring of vibration signal, we design a system based on wavelet transform and Elman neural network. And in this system, wavelet transform has been considered to be used for digital signal procession and for feature extraction, while using the pattern recognition ability of neural network to determine the status of electrical motor work.
This paper analyzes the process at work in the motor common working conditions, including the shell burst, the base loose, the rotor does not work in the right place, these three common failure modes, and normal working conditions, collected by two different vibration sensor signals. A set of training the neural network is used as a sample signal; the other group used the trained neural network performance testing, as test signals. The signal eigenvectors of vibration signal for the training were extracted by wavelet packet by the, and then used for neural network training. Feature vector extraction also has been conducted to the test signal, and then passed them through the trained neural network to diagnose the situation of electrical motor work.
In this paper, the design of the diagnostic system has been simulated on Matlab platform, and the system simulation is able to verify the validity and accuracy of the system. Test result is consistent with the actual test signals corresponding to different states. And from the result, we can see the diagnosis system based on wavelet transform and Elman neural network in this article is able to conduct an effective diagnosis for the working status of motor.
Finally, after review the whole system design, the outlook of future work has been depicted.
本文在总结了传统的电机故障诊断方法的基础上,通过对电机工作中振动信号的采集与监测以及对电机工作故障的分析,设计了一种基于小波包变换与Elman人工神经网络的电机故障诊断系统,通过小波包变换对采集数据进行信号处理与特征值提取,并利用神经网络的模式识别能力对电机的工作状况进行判定。
本文研究分析了电机在工作过程中常见的工作状况,并针对外壳破裂、基座松脱、转子不对中等三种常见工作故障模式以及正常工作状况通过传感器采集两组不同的振动信号。一组用于对神经网络进行训练,作为样本信号；另一组用于对训练好的神经网络进行性能测试,作为测试信号。对于用于训练学习的振动信号用小波包变换的方法对信号进行特征值提取得到信号的特征向量,并对神经网络系统进行训练。对测试信号进行同样的特征向量提取,并通过训练好的神经网络对电机的工作状况进行诊断。
本文对上述所设计的诊断系统在Matlab平台上进行了系统仿真,验证了算法的有效性和准确性。测试结果符合实际测试信号对应的不同状态,结果证明了本文中所设计的基于小波包变换和Elman人工神经网络电机故障诊断系统可以有效的对直流电动机在工作过程中发生的故障做出有效的诊断。
最后本文对设计的诊断系统进行了必要的总结,并对课题未来延伸的研究方向进行了展望。
With the rapid development of modern industry, particularly after production line is widely used in industrial, the motor has become an important base of modern industrial technology. Modern motor design has not only been driven by the way how to improve their capacity, while the security, stability and reliability have also become the significant aspect cannot be ignored. Therefore, how to effectively recognize the failure occurred during the industrial production will be an important influence for the stable and orderly conduct of the process.
In this work, based on the summary of traditional motor fault diagnosis methods and the analysis of acquisition and monitoring of vibration signal, we design a system based on wavelet transform and Elman neural network. And in this system, wavelet transform has been considered to be used for digital signal procession and for feature extraction, while using the pattern recognition ability of neural network to determine the status of electrical motor work.
This paper analyzes the process at work in the motor common working conditions, including the shell burst, the base loose, the rotor does not work in the right place, these three common failure modes, and normal working conditions, collected by two different vibration sensor signals. A set of training the neural network is used as a sample signal; the other group used the trained neural network performance testing, as test signals. The signal eigenvectors of vibration signal for the training were extracted by wavelet packet by the, and then used for neural network training. Feature vector extraction also has been conducted to the test signal, and then passed them through the trained neural network to diagnose the situation of electrical motor work.
In this paper, the design of the diagnostic system has been simulated on Matlab platform, and the system simulation is able to verify the validity and accuracy of the system. Test result is consistent with the actual test signals corresponding to different states. And from the result, we can see the diagnosis system based on wavelet transform and Elman neural network in this article is able to conduct an effective diagnosis for the working status of motor.
Finally, after review the whole system design, the outlook of future work has been depicted.
引文
[1]R.C. Luo, M.G. Kay, "Multi-sensor integration and fusion in intelligent systems", IEEE Trans. Systems Man Cybernet.19 (1989) 901-931.
[2]A. Noori-khajavi, R. Komanduri, "on multisensor approach to drill wear monitoring", Ann.CIRP 42(1993) 71-74.
[3]I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci, "Wireless Sensor Networks:A Survey", Computer Networks, Computer Networks Journal, vol.38,2002, pp.102-105.
[4]F. Hlawatsch and G F. Boudreaux-Bartels, "Linear and quadratic timekequency signal representations," IEEE Signal Processing Mag., vol.9, pp.21-67, Apr.1992.
[5]0. Rioul and M. Vetterli, "Wavelets and signal processing," IEEE Signal Processing Mag., vol.8, pp.14-38, Oct.1991.
[6]Li, C. J.,& Ma, J, "Wavelet decomposition of vibrations for detection of bearing-localized defects." NDT & E International,30,143-149.1997
[7]Sung, C. K., Tai, H. M., & Chen, C. W. "Locating defects of a gear system by the technique of wavelet transform". Mechanism and Machine Theory,35,1169-1182.2000
[8]Zheng, H., Li, Z.,& Chen, X. (2002). "Gear fault diagnosis based on continuous wavelet transform". Mechanical System and Signal Processing,16,447-457.
[9]D. Gabor, "Theory of communication", Journal of I.E.E.93 pp 429-441,1946.
[10]Y. Meyer, Wavelets, Ed. J.M. Combes et al., Springer Verlag, Berlin, p.21,1989.
[II]S.GMallat,A theory of multire solution signal decomposition:the wavelet representation,IEEE Transactionson Pattern and Machine Intelligence 11 (7) 2-9 (1989)674-693. [64]
[12]I.N.Tansel,C.Mekdeci,O.Rodriguez,B.Uragun,Monitoring drill conditions with wavelet based encoding and neural networks,International Journal of Machine Tools and Manufacture 33 (1993)559-575.
[13]I.N.Tansel,C.Mekdeci,C.McLaughlin,Detection of tool failure in end milling with wavelet transformations and neural networks(WT-NN),International Journal of Machine Tools and Manufacture 35 (August) (1995)1137-1147. [14] S.GMallat,A Wavelet Tour of Signal Processing,seconded., AcademicPress, NewYork,1999.
[15]M.V.Wickerhauser,R.R.Coifman,Entropy based methods for best basis selection,IEEE Transactions on Information Theory 38(2) (1992) 719-746.
[16]0. Rioul and M. Vetterli, "Wavelets and signal processing," IEEE Signal Processing Mag., vol. 8, pp.14-38, Oct.1991.
[17]I.E. Frank, Chemometri. Intel. Lab. Syst.27 (1995) 1.
[18]J.L. Elman, "Finding structure in time," Cognitive Science.vol.14, pp.179-211,1990.
[19]沈轶,廖晓昕.离散神经网络的全局收敛性[J].系统工程与电子技术.1999：21(11)：50-51
[20]吴微,徐东坡,李正学.Elman人网络梯度学习法的收敛性[J].应用数学和力学,第29卷第9期.2008.9.15
[21]黄聪明,李志坚.基于改进的递归神经网络的化工动态系统建模[J].北京理工大学学报,2004,24(7)：596-599.
[22]范燕,中东日,陈义俊,等.基于改进ELMAN网络的非线性预测控制[J].河南科技大学学报,2007,28(1)：41-45.
[23]郑成兴.网络流量预测方法和实际预测分析[J].计算机工程应用,2006,42(23)：129-130.
[24]康海英,栾军英,郑海起,等.基于小波包变换和BP网络的齿轮箱故障诊断[J].军械工程学院学报,2005,4：26-28.
[25]沈标正.电机故障诊断技术[M].北京：机械工业出版社,1996.
[26]高景德,王祥珩,李发海.交流电机及其系统的分析[M].北京：清华大学出版社,1993.
[27]梁霖.基于电流法的鼠笼异步电动机故障特征提取及其自动诊断[D].西安：西安交通大学,2001.
[28]王杰,闫东伟.提高预测精度的ELMAN和SOM神经网络组合[J].系统工程与电子技术,2004,26(12)：1943-1945.
[29]李辉,郑海起,杨绍普.基于角域平均和连续小波变换的齿轮故障诊断研究[J].振动与冲击,2007,26(11)：16-19
[30]史成江.直流大电机在线监测与故障诊断专家系统[D].湖北：武汉科技大学,2005
[31]张植保.电机原理及其运行与维护[M].北京：化学工业出版社,2005
[32]肖方.直流电机在线监测与故障诊断的应用矶究[D].湖北：武汉科技大学,2003
[33]姜洪开,王仲生,何正嘉.基于改进第2代小波算法的发电机组碰摩故障特征提取[J].中国电机工程学报,2008,28(8)：127-131
[34]赵佰亭,陈希军,曾庆双.基于小波变换的精密测试转台测角系统的故障诊断[J].中国惯性技术学报,2007,15(5)：630-634
[35]武建军,马振利,秦瑞胜等.小波技术在车载发动机泵机组故障诊断中的应用[J].机床与液压,2007,35(11)：183-185
[36]周伟,桂林,周林.MATLAB小波分析高级技术[M].西安：西安电子科技大学出版社,200552[37]孙延奎.小波分析及其应用[M].北京：机械工业出版社,2005
[38]Zhiyong Luo,Zhongke Shi.Wavelet Neural Network Method for Fault Diagnosis of PUSH-PULL Circuits[C].Proceedings of the Fourth International Conference on Machine Learning and Cybernetics, Guangzhou,2005,3327-3332
[39]徐晨,赵瑞珍,甘小冰.小波分析·应用算法[M].北京：科学出版社,2004
[40]边威.小波基的选取与构造方法讨论[D].辽宁：东北师范大学,2007
[41]Liu Hongxing,Li Jian,Zhao Ying.Improved singular value decomposition technique for detecting and extracting periodic impulse component in a vibration signal[J].Chinese Journal of Mechanical Engineering(English Edition),2004,17(3):340-346
[42]毋文峰,王汉功,陈小虎.基于小波包能量谱-神经网络的液压泵故障诊断[J].液压与气动,2006,30(12)：85-88
[2]A. Noori-khajavi, R. Komanduri, "on multisensor approach to drill wear monitoring", Ann.CIRP 42(1993) 71-74.
[3]I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci, "Wireless Sensor Networks:A Survey", Computer Networks, Computer Networks Journal, vol.38,2002, pp.102-105.
[4]F. Hlawatsch and G F. Boudreaux-Bartels, "Linear and quadratic timekequency signal representations," IEEE Signal Processing Mag., vol.9, pp.21-67, Apr.1992.
[5]0. Rioul and M. Vetterli, "Wavelets and signal processing," IEEE Signal Processing Mag., vol.8, pp.14-38, Oct.1991.
[6]Li, C. J.,& Ma, J, "Wavelet decomposition of vibrations for detection of bearing-localized defects." NDT & E International,30,143-149.1997
[7]Sung, C. K., Tai, H. M., & Chen, C. W. "Locating defects of a gear system by the technique of wavelet transform". Mechanism and Machine Theory,35,1169-1182.2000
[8]Zheng, H., Li, Z.,& Chen, X. (2002). "Gear fault diagnosis based on continuous wavelet transform". Mechanical System and Signal Processing,16,447-457.
[9]D. Gabor, "Theory of communication", Journal of I.E.E.93 pp 429-441,1946.
[10]Y. Meyer, Wavelets, Ed. J.M. Combes et al., Springer Verlag, Berlin, p.21,1989.
[II]S.GMallat,A theory of multire solution signal decomposition:the wavelet representation,IEEE Transactionson Pattern and Machine Intelligence 11 (7) 2-9 (1989)674-693. [64]
[12]I.N.Tansel,C.Mekdeci,O.Rodriguez,B.Uragun,Monitoring drill conditions with wavelet based encoding and neural networks,International Journal of Machine Tools and Manufacture 33 (1993)559-575.
[13]I.N.Tansel,C.Mekdeci,C.McLaughlin,Detection of tool failure in end milling with wavelet transformations and neural networks(WT-NN),International Journal of Machine Tools and Manufacture 35 (August) (1995)1137-1147. [14] S.GMallat,A Wavelet Tour of Signal Processing,seconded., AcademicPress, NewYork,1999.
[15]M.V.Wickerhauser,R.R.Coifman,Entropy based methods for best basis selection,IEEE Transactions on Information Theory 38(2) (1992) 719-746.
[16]0. Rioul and M. Vetterli, "Wavelets and signal processing," IEEE Signal Processing Mag., vol. 8, pp.14-38, Oct.1991.
[17]I.E. Frank, Chemometri. Intel. Lab. Syst.27 (1995) 1.
[18]J.L. Elman, "Finding structure in time," Cognitive Science.vol.14, pp.179-211,1990.
[19]沈轶,廖晓昕.离散神经网络的全局收敛性[J].系统工程与电子技术.1999：21(11)：50-51
[20]吴微,徐东坡,李正学.Elman人网络梯度学习法的收敛性[J].应用数学和力学,第29卷第9期.2008.9.15
[21]黄聪明,李志坚.基于改进的递归神经网络的化工动态系统建模[J].北京理工大学学报,2004,24(7)：596-599.
[22]范燕,中东日,陈义俊,等.基于改进ELMAN网络的非线性预测控制[J].河南科技大学学报,2007,28(1)：41-45.
[23]郑成兴.网络流量预测方法和实际预测分析[J].计算机工程应用,2006,42(23)：129-130.
[24]康海英,栾军英,郑海起,等.基于小波包变换和BP网络的齿轮箱故障诊断[J].军械工程学院学报,2005,4：26-28.
[25]沈标正.电机故障诊断技术[M].北京：机械工业出版社,1996.
[26]高景德,王祥珩,李发海.交流电机及其系统的分析[M].北京：清华大学出版社,1993.
[27]梁霖.基于电流法的鼠笼异步电动机故障特征提取及其自动诊断[D].西安：西安交通大学,2001.
[28]王杰,闫东伟.提高预测精度的ELMAN和SOM神经网络组合[J].系统工程与电子技术,2004,26(12)：1943-1945.
[29]李辉,郑海起,杨绍普.基于角域平均和连续小波变换的齿轮故障诊断研究[J].振动与冲击,2007,26(11)：16-19
[30]史成江.直流大电机在线监测与故障诊断专家系统[D].湖北：武汉科技大学,2005
[31]张植保.电机原理及其运行与维护[M].北京：化学工业出版社,2005
[32]肖方.直流电机在线监测与故障诊断的应用矶究[D].湖北：武汉科技大学,2003
[33]姜洪开,王仲生,何正嘉.基于改进第2代小波算法的发电机组碰摩故障特征提取[J].中国电机工程学报,2008,28(8)：127-131
[34]赵佰亭,陈希军,曾庆双.基于小波变换的精密测试转台测角系统的故障诊断[J].中国惯性技术学报,2007,15(5)：630-634
[35]武建军,马振利,秦瑞胜等.小波技术在车载发动机泵机组故障诊断中的应用[J].机床与液压,2007,35(11)：183-185
[36]周伟,桂林,周林.MATLAB小波分析高级技术[M].西安：西安电子科技大学出版社,200552[37]孙延奎.小波分析及其应用[M].北京：机械工业出版社,2005
[38]Zhiyong Luo,Zhongke Shi.Wavelet Neural Network Method for Fault Diagnosis of PUSH-PULL Circuits[C].Proceedings of the Fourth International Conference on Machine Learning and Cybernetics, Guangzhou,2005,3327-3332
[39]徐晨,赵瑞珍,甘小冰.小波分析·应用算法[M].北京：科学出版社,2004
[40]边威.小波基的选取与构造方法讨论[D].辽宁：东北师范大学,2007
[41]Liu Hongxing,Li Jian,Zhao Ying.Improved singular value decomposition technique for detecting and extracting periodic impulse component in a vibration signal[J].Chinese Journal of Mechanical Engineering(English Edition),2004,17(3):340-346
[42]毋文峰,王汉功,陈小虎.基于小波包能量谱-神经网络的液压泵故障诊断[J].液压与气动,2006,30(12)：85-88