隐马尔科夫模型及其在机械故障模式识别中的应用
|
摘要
隐马尔科夫模型(Hidden Markov Model,简称作HMM)是一种新的模式识别技术,其基本方法是通过对训练信号进行特征值提取和标量量化,建立具有相应状态数和观测值数的隐Markov模型,然后利用该模型计算待诊断信号与训练信号的相似概率,根据相似概率的差异判断信号状态的变化,达到信号模式分类的目的。本文的主要研究工作是利用HMM进行机械设备运行状态识别和故障诊断,包括三项研究内容:1)在深入学习HMM理论基础上,探讨HMM的技术实现方法,编制Matlab环境下的HMM实现程序;2)研究振动信号特征提取方法,开发基于LabVIEW的希尔伯特变换和倒频谱分析程序;3)应用HMM对于汽轮机和齿轮等两种典型旋转机械设备运行状态进行分析识别。对于汽轮机设备,以其振动信号频谱中基频处的幅值作为HMM训练的特征值,建立汽轮机升负荷过程的HMM,进行设备状态变化分析;对于齿轮箱设备,建立运行过程振动信号的HMM模型,根据相似概率的变化识别齿根裂纹故障的生成及发展趋势,两个应用案例都给出满意的结果。
Hidden Markov Model (HMM) is a new technique in pattern recognition. By extracting and vector quantizing the features of training signal, Hidden Markov Mode with according state number and observation number can be established, then the similarity probability of the unknown signal can be calculated through HMM. By comparing the similarity probability, pattern of the signal can be recognized. In this work, HMM is applied in the pattern recognition and fault diagnosis of mechanical equipment, The mean works include three parts: 1) The realization approach of HMM for fault diagnosis was discussed and the program of HMM was developed with Matlab; 2) The methods of feature extraction from vibration signal, especially cepstrum and envelop method, was studied, and according program was developed with LabVIEW; 3) Two typical rotating machine(turbine and gearbox) with different signal characteristics were used to improve the effectiveness of HMM for identification and classification of machine condition. For the turbine, the amplitude at the basic frequency is extracted as the features for HMM training, for the gear box, HMM is established to recognize the development of gear crack. The results of these two examples are all satisfied.
Hidden Markov Model (HMM) is a new technique in pattern recognition. By extracting and vector quantizing the features of training signal, Hidden Markov Mode with according state number and observation number can be established, then the similarity probability of the unknown signal can be calculated through HMM. By comparing the similarity probability, pattern of the signal can be recognized. In this work, HMM is applied in the pattern recognition and fault diagnosis of mechanical equipment, The mean works include three parts: 1) The realization approach of HMM for fault diagnosis was discussed and the program of HMM was developed with Matlab; 2) The methods of feature extraction from vibration signal, especially cepstrum and envelop method, was studied, and according program was developed with LabVIEW; 3) Two typical rotating machine(turbine and gearbox) with different signal characteristics were used to improve the effectiveness of HMM for identification and classification of machine condition. For the turbine, the amplitude at the basic frequency is extracted as the features for HMM training, for the gear box, HMM is established to recognize the development of gear crack. The results of these two examples are all satisfied.
引文
[1] Hasan Ocak. Fault detection, diagnosis and prognosis of rolling element bearings: frequency domain methods and Hidden Markov Modeling[J]. ProQuest information and learning company (2004)
[2] Kee-Choon Kwon, Jin-Hyung Kim. Accident identification in nuclear power plants using Hidden Markov Models[J]. Engineering Applications of Artificial Intelligence 12 (1999) p491-501
[3] James C. Wong, Karen A. McDonald, Ahmet Palazoglu. Classification of abnormal plant operation using multiple process variable trends[J]. Journal of Process Control 11 (2001) 409-418
[4] Jong Min Lee, Seung-Jong Kim, Yoha Hwang, Chang-Seop Song. Diagnosis of mechanical fault signals using continuous Hidden Markov Model[J]. Journal of Sound and Vibration 276 (2004) 1065-1080
[5] Zhinong Li, Zhaotong, Yongyong He, Chu Fulei. Hidden Markov model-based diagnostics method in speed-up and speed-down process for rotating machinery[J]. Mechanical Systems and Signal Processing 19 (2005) 329-339
[6] Oded Netzer. A Hidden Markov Model of customer relationsip dynamics [7].Engineering Applications of Artificial Intelligence 12 (1999) 491-501
[7] Jeff Bilmes. What HMMs Can Do[J]. UWEE Technical Report . Number UWEETR-2002-2003
[8] Huseyin Metin Ertunc,Cuneyt Oysu. Drill wear monitoring using cutting force signals [J].Mechatronics 14 (2004) 533-548
[9] S.E.Levinson, L.R.Rabiner, M.M.Sondhi. An introduction to the application of the theory of probabilistic functions of a Markov process to automatic speech recognition[J]. The Bell System Technical Journal, (1983) 1035-1073
[10] S J Cox. Hidden Markov Models for automatic speech recognition: theory and application[J]. In C. Wheddon and R.Linggard, editors, Speech and Language Processing, p209-230,1990
[11] 2003基于隐Markov模型的旋转机械故障诊断系统的研究.NH
[12] 叶福军.HMM和SVM相结合的音频自动分类[J].高性能计算技术:2005, vol.172
[13] 李培玉,邹福星. HMM在钢水连铸下渣振动检测方法中的应用[J]. 冶金自动化 2005年第1期
[14] 芮挺,沈春林,张金林. 基于HMM的基健特征用户身份认证[J]. 计算机工程:2005,vol.31 No.2
[15] 王崇文,李见为,陈为民. HMM和SVM相结合的指纹分类方法[J]. 电子与信息学报: 2003,vol.25 No.11
[16] Xu Xianghua, Zhu Jie, Guo Qiang. Speaker-independent speech recognition based on HMM state-restructuring method [J]. Journal of Southeast University(English Edition ): 2004, Vol.20 No.4
[17] 汪鹏,刘加,刘润生.基于离散HMM的非特定人关键词提取语音识别系统[J]. 吉林大学学报(理学版): 2003,vol.41 No.3 347-351
[18] 张梅.隐Markov模型的基本原理及其在基因识别中的应用[J]. 山西科技大学学报: 2003,vol.21 No.4
[19] 胡广书. 数字信号处理——理论、算法与实现[M]. 北京:清华大学出版社,1997
[20] 杨志伊,郑文. 设备状态监测与故障诊断[M]. 北京:化学工业出版社,2003
[21] 盛兆顺,尹琦岭. 设备状态监测与故障诊断技术及应用[M]. 北京:中国计划出版社,2006
[22] 周浩敏. 信号处理技术基础[M]. 北京:北京航空航天大学出版社,2002
[23] 丁康,李巍华,朱小勇. 齿轮及齿轮箱故障诊断实用技术[M]. 北京:机械工业出版社,2005
[24] 王炳锡,屈丹,彭煊. 实用语音识别技术[M]. 北京:国防工业出版社,2005
[25] 于润伟. Matlab基础及应用[M]. 北京:机械工业出版社,2005
[26] 杨乐平,李海涛,赵勇,杨磊,安雪滢. LabVIEW高级程序设计[M]. 北京:清华大学出版社,2004
[27] 冯长建. HMM动态模式识别理论、方法及在机械故障诊断中的应用[M]. 北京:化学工业出版社,2003
[28] Matrouf. D, Gauvain, J L. Using AR HMM state-dependent filtering for speech enhancement[J].Acoustics, Speech, and Signal Processing , 1999. Proceeding . 1999 IEEE International Conference, 1999,2: 785-788
[29] LAWRENCE R. RABINER. A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition[J]. Proceedings of the IEEE, Vol. 77, No.2, 1989
[30] 飞思科技产品研发中心. 神经网络理论与MATLAB7实现[M]. 北京:电子工业出版,2005,234-250
[31] Ray Ruichong Zhang, Lance VanDemark, Jianwen Liang, Yuxian Hu. On estimating site damping with soil non-linearity from earthquake recordings. International Journal of Non-Linear Mechanics 39 (2004) 1501-1517
[32] Marcus Datig, Torsten Schlurmann. Performance and limitations of the Hilbert-Huangtransformation (HHT) with an application to irregular water waves. Ocean Engineering 31 (2004) 1783–1834
[33] 张帆. 信号处理中的解调分析技术及齿轮箱典型故障特征的提取. 汕头大学硕士学位论文 2003
[34] 赵海鸣,卜英勇,何学文.旋转机械在线状态监测与故障诊断[J]. 机电工程: 2003,vol.23 No.8
[35] 白士红,孙斌.包络方法诊断齿轮箱故障[J]. 沈阳航空工业学院学报: 2000,vol.17 No.4
[36] 胡劲松,杨世锡.基于HHT的旋转机械故障诊断方法研究[J]. 动力工程: 2004,vol.24 No.6
[37] 李四信,韦岗.连续型隐马尔可夫模型(HMM)参数与语音识别[J]. 华中师范大学学报: 1998,vol.32 No.1
[2] Kee-Choon Kwon, Jin-Hyung Kim. Accident identification in nuclear power plants using Hidden Markov Models[J]. Engineering Applications of Artificial Intelligence 12 (1999) p491-501
[3] James C. Wong, Karen A. McDonald, Ahmet Palazoglu. Classification of abnormal plant operation using multiple process variable trends[J]. Journal of Process Control 11 (2001) 409-418
[4] Jong Min Lee, Seung-Jong Kim, Yoha Hwang, Chang-Seop Song. Diagnosis of mechanical fault signals using continuous Hidden Markov Model[J]. Journal of Sound and Vibration 276 (2004) 1065-1080
[5] Zhinong Li, Zhaotong, Yongyong He, Chu Fulei. Hidden Markov model-based diagnostics method in speed-up and speed-down process for rotating machinery[J]. Mechanical Systems and Signal Processing 19 (2005) 329-339
[6] Oded Netzer. A Hidden Markov Model of customer relationsip dynamics [7].Engineering Applications of Artificial Intelligence 12 (1999) 491-501
[7] Jeff Bilmes. What HMMs Can Do[J]. UWEE Technical Report . Number UWEETR-2002-2003
[8] Huseyin Metin Ertunc,Cuneyt Oysu. Drill wear monitoring using cutting force signals [J].Mechatronics 14 (2004) 533-548
[9] S.E.Levinson, L.R.Rabiner, M.M.Sondhi. An introduction to the application of the theory of probabilistic functions of a Markov process to automatic speech recognition[J]. The Bell System Technical Journal, (1983) 1035-1073
[10] S J Cox. Hidden Markov Models for automatic speech recognition: theory and application[J]. In C. Wheddon and R.Linggard, editors, Speech and Language Processing, p209-230,1990
[11] 2003基于隐Markov模型的旋转机械故障诊断系统的研究.NH
[12] 叶福军.HMM和SVM相结合的音频自动分类[J].高性能计算技术:2005, vol.172
[13] 李培玉,邹福星. HMM在钢水连铸下渣振动检测方法中的应用[J]. 冶金自动化 2005年第1期
[14] 芮挺,沈春林,张金林. 基于HMM的基健特征用户身份认证[J]. 计算机工程:2005,vol.31 No.2
[15] 王崇文,李见为,陈为民. HMM和SVM相结合的指纹分类方法[J]. 电子与信息学报: 2003,vol.25 No.11
[16] Xu Xianghua, Zhu Jie, Guo Qiang. Speaker-independent speech recognition based on HMM state-restructuring method [J]. Journal of Southeast University(English Edition ): 2004, Vol.20 No.4
[17] 汪鹏,刘加,刘润生.基于离散HMM的非特定人关键词提取语音识别系统[J]. 吉林大学学报(理学版): 2003,vol.41 No.3 347-351
[18] 张梅.隐Markov模型的基本原理及其在基因识别中的应用[J]. 山西科技大学学报: 2003,vol.21 No.4
[19] 胡广书. 数字信号处理——理论、算法与实现[M]. 北京:清华大学出版社,1997
[20] 杨志伊,郑文. 设备状态监测与故障诊断[M]. 北京:化学工业出版社,2003
[21] 盛兆顺,尹琦岭. 设备状态监测与故障诊断技术及应用[M]. 北京:中国计划出版社,2006
[22] 周浩敏. 信号处理技术基础[M]. 北京:北京航空航天大学出版社,2002
[23] 丁康,李巍华,朱小勇. 齿轮及齿轮箱故障诊断实用技术[M]. 北京:机械工业出版社,2005
[24] 王炳锡,屈丹,彭煊. 实用语音识别技术[M]. 北京:国防工业出版社,2005
[25] 于润伟. Matlab基础及应用[M]. 北京:机械工业出版社,2005
[26] 杨乐平,李海涛,赵勇,杨磊,安雪滢. LabVIEW高级程序设计[M]. 北京:清华大学出版社,2004
[27] 冯长建. HMM动态模式识别理论、方法及在机械故障诊断中的应用[M]. 北京:化学工业出版社,2003
[28] Matrouf. D, Gauvain, J L. Using AR HMM state-dependent filtering for speech enhancement[J].Acoustics, Speech, and Signal Processing , 1999. Proceeding . 1999 IEEE International Conference, 1999,2: 785-788
[29] LAWRENCE R. RABINER. A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition[J]. Proceedings of the IEEE, Vol. 77, No.2, 1989
[30] 飞思科技产品研发中心. 神经网络理论与MATLAB7实现[M]. 北京:电子工业出版,2005,234-250
[31] Ray Ruichong Zhang, Lance VanDemark, Jianwen Liang, Yuxian Hu. On estimating site damping with soil non-linearity from earthquake recordings. International Journal of Non-Linear Mechanics 39 (2004) 1501-1517
[32] Marcus Datig, Torsten Schlurmann. Performance and limitations of the Hilbert-Huangtransformation (HHT) with an application to irregular water waves. Ocean Engineering 31 (2004) 1783–1834
[33] 张帆. 信号处理中的解调分析技术及齿轮箱典型故障特征的提取. 汕头大学硕士学位论文 2003
[34] 赵海鸣,卜英勇,何学文.旋转机械在线状态监测与故障诊断[J]. 机电工程: 2003,vol.23 No.8
[35] 白士红,孙斌.包络方法诊断齿轮箱故障[J]. 沈阳航空工业学院学报: 2000,vol.17 No.4
[36] 胡劲松,杨世锡.基于HHT的旋转机械故障诊断方法研究[J]. 动力工程: 2004,vol.24 No.6
[37] 李四信,韦岗.连续型隐马尔可夫模型(HMM)参数与语音识别[J]. 华中师范大学学报: 1998,vol.32 No.1