基于BP神经网络的发动机异响模式识别
|
摘要
传统的异响类型诊断方法大多是以领域专家和操作者的经验知识为核心,存在知识获取困难、推理效率低下、自适应能力差等不足。同时由于异响类型征兆和异响类型之间常存在着复杂的非线性关系,使得诊断系统的数学模型很难获取。然而,人工神经网络以其分布式并行处理、自适应、自学习、联想记忆以及非线性映射等优点,为解决这一问题开辟了新途径。
本文以JS某型摩托车发动机为研究对象。以连续小波提取发动机声信号的能量特征作为输入特征值,建立相应的BP神经网络运用标准BP算法对几种异响类型进行分类,实现模式识别的目的。然而由于建立的神经网络采用标准BP算法时存在对样本的输入顺序敏感、收敛速度缓慢、易陷入局部极小值等缺陷,为了克服这些不足需对标准BP算法进行改进。本文针对标准BP算法存在的不足,分别采用打乱样本输入顺序、添加附加动量因子、学习率自适应调整和基于Levenberg-Marquardt法进行改进。这几种改进措施分别对应了加动量项BP算法、附加动量—自适应学习率BP算法及基于Levenberg-Marquardt算法,本文对上述三种改进算法及标准BP算法进行分析对比。
最终,通过对比几种算法和不同网络结构的诊断速度和诊断准确率,得出了最适合的神经网络结构为三层网络:输入层单元数为11、隐层单元数为20、输出层单元数为2;输入层—隐含层传递函数为tansig、隐含层—输出层传递函数为logsig、训练函数为trainlm;最适合的算法为基于Levenberg-Marquardt算法。最后,由于编写相应算法的M文件有些繁杂,本文为了提高人机互动性,应用MATLAB的图形用户界面(GUI)设计开发了发动机异响类型诊断系统。
For the diagnosis in abnormal sound engines, the traditional methods of the pattern recognition usually based on experience and knowledge of experts and operators. There are many defects in the traditional methods, which contains difficulties in knowledge acquisition, inference inefficient and low adaptive ability. Meanwhile, it is difficult to acquire mathematics model of the diagnostic system, because the complex nonlinear relationship existed between characteristics and types of abnormal sounds. The artificial neural network provides a new solution to this problem due to its unique advantages, such as parallel distributed processing, self-adaptation, self-learning, associational memory and so on.
Based on a type of JS motorcycle and using energy features of acoustic signals as the inputs of network, which was extracted by continuous wavelet, a BP neural network is established in this paper using standard BP algorithm in order to achieve the purpose of the classification of several types of abnormal sounds. However, the BP neural network using standard BP algorithm has some defects, for example, it’s sensitive to the order of the sample data, slow convergence, and easy to get into the local minimum. Improvements are introduced in this paper, they are disrupting the order of sample input, adding additional momentum, adaptive learning rate and improved BP algorithm based on Levenberg-Marquardt. The corresponding algorithms of these improvements are the adaptive learning rate BP algorithm, the BP algorithm combined additional momentum with adaptive learning rate and improved algorithms based on Levenberg-Marquardt. This paper makes a detailed analysis about those algorithms which is mentioned above, and the effect of different parameter in the improved algorithms.
Finally, comparing the speed of convergence and the accuracy of diagnosis, this paper gets the optimization algorithm and network. The optimization network is three layers network structure: input layer has eleven neurons, the hidden layer has twenty neurons and output layer has two neurons. The optimization algorithm is the improved algorithm based on Levenberg-Marquardt. In the end, the front panel of the diagnosis system is designed by MATLAB graphical user interface (GUI).
本文以JS某型摩托车发动机为研究对象。以连续小波提取发动机声信号的能量特征作为输入特征值,建立相应的BP神经网络运用标准BP算法对几种异响类型进行分类,实现模式识别的目的。然而由于建立的神经网络采用标准BP算法时存在对样本的输入顺序敏感、收敛速度缓慢、易陷入局部极小值等缺陷,为了克服这些不足需对标准BP算法进行改进。本文针对标准BP算法存在的不足,分别采用打乱样本输入顺序、添加附加动量因子、学习率自适应调整和基于Levenberg-Marquardt法进行改进。这几种改进措施分别对应了加动量项BP算法、附加动量—自适应学习率BP算法及基于Levenberg-Marquardt算法,本文对上述三种改进算法及标准BP算法进行分析对比。
最终,通过对比几种算法和不同网络结构的诊断速度和诊断准确率,得出了最适合的神经网络结构为三层网络:输入层单元数为11、隐层单元数为20、输出层单元数为2;输入层—隐含层传递函数为tansig、隐含层—输出层传递函数为logsig、训练函数为trainlm;最适合的算法为基于Levenberg-Marquardt算法。最后,由于编写相应算法的M文件有些繁杂,本文为了提高人机互动性,应用MATLAB的图形用户界面(GUI)设计开发了发动机异响类型诊断系统。
For the diagnosis in abnormal sound engines, the traditional methods of the pattern recognition usually based on experience and knowledge of experts and operators. There are many defects in the traditional methods, which contains difficulties in knowledge acquisition, inference inefficient and low adaptive ability. Meanwhile, it is difficult to acquire mathematics model of the diagnostic system, because the complex nonlinear relationship existed between characteristics and types of abnormal sounds. The artificial neural network provides a new solution to this problem due to its unique advantages, such as parallel distributed processing, self-adaptation, self-learning, associational memory and so on.
Based on a type of JS motorcycle and using energy features of acoustic signals as the inputs of network, which was extracted by continuous wavelet, a BP neural network is established in this paper using standard BP algorithm in order to achieve the purpose of the classification of several types of abnormal sounds. However, the BP neural network using standard BP algorithm has some defects, for example, it’s sensitive to the order of the sample data, slow convergence, and easy to get into the local minimum. Improvements are introduced in this paper, they are disrupting the order of sample input, adding additional momentum, adaptive learning rate and improved BP algorithm based on Levenberg-Marquardt. The corresponding algorithms of these improvements are the adaptive learning rate BP algorithm, the BP algorithm combined additional momentum with adaptive learning rate and improved algorithms based on Levenberg-Marquardt. This paper makes a detailed analysis about those algorithms which is mentioned above, and the effect of different parameter in the improved algorithms.
Finally, comparing the speed of convergence and the accuracy of diagnosis, this paper gets the optimization algorithm and network. The optimization network is three layers network structure: input layer has eleven neurons, the hidden layer has twenty neurons and output layer has two neurons. The optimization algorithm is the improved algorithm based on Levenberg-Marquardt. In the end, the front panel of the diagnosis system is designed by MATLAB graphical user interface (GUI).
引文
[1]殷勤业等.模式识别[M].北京:机械工业出版社,1999.
[2]张德丰等.MATLAB神经网络应用设计[M].机械工业出版社,2009.
[3]任志英.基于声信号技术的发动机故障诊断系统研究[D].福州大学硕士学位论文,2005.
[4] D. W. Thomas, B. R. Wilkins: The analysis of vehicle sounds for recognition, Pattern Recognition[J].1972.l4, 379-389.
[5] J.C. Webster, M. M. Woodhead, A. Carpenter: Identifying diesel engine sounds [J]. Journal of Sound Vibration,Vo1.9, No.2,2 41-246,1969.
[6] Richard H.L yon. Machinery Noise and Diagnostics [M].LONDON:Buterworth,1987.
[7] S.K.Lee, P.R.White. Higher-order and time-frequency analysis and its application to fault detection in rotating machinery[J].Mechanical Systems and Signal Processing,1997, 11,(4):637-650.
[8]孙宗新.基于混沌理论的发动机异响信号提取研究[D].吉林大学硕士学位论文,2007.6.
[9]舒大文等.用振动和噪声信号对汽车变速箱齿轮的研究[J].昆明理工大学学报,1997,4.
[10]李建秋,欧阳明高.汽车发动机瞬时转速测量中的噪声分析及补偿方法[J].中国机械工程,2000,7.
[11]刘福.基于声音的发动机诊断方法研究与实践[D].东北大学硕士学位论文,2006,6.
[12] Lin Jing,Qu Liang-sheng,Feature extraction based on Morlet wavelet and its application for mechanical fault diagnosis[J].Journal of Vibration,2000.234(1):135-148.
[13] Liu Baoling,Shih-fu,Meng Qingfeng,Machinery diagnosis based on wavelet packets[J].Journal of Vibration and Control,1997,3(1):5-7.
[14] Yen,Gary G,Lin Kuo-Chung,Conditional health monitoring using vibration signatures[C].Proceedings of the IEEE Conference on Decision and Control,1999,5:4493-4498.
[15] Yen,Gary G,Lin Kuo-Chung,Wavelet packet feature extraction for vibration monitoring[C].IEEE Transactions on Industrial Electronics,2000,47(3):650-667
[16] Hambaba A,Huff E,Multiresolution error detection on early fatigur cracks in gears[C].IEEE Aerospace Conference Proceeding,2000,6:367-373
[17]赵纪元,何正嘉,孟庆丰等.小波包—自回归谱分析在振动诊断中的应用[J].振动工作学报,1995,8(3):198-203.
[18]葛郢汉.发动机异响的分析与诊断[J].汽车博士,2007,(2):37-38.
[19]周志新.摩托车顶杆式发动机异响分析[J].摩托车,2003:49-50.
[20]刘永志,王建中.摩托车发动机常见异响的判断及预防[J].摩托车,2005:33-36.
[21]魏少华,陈效华,常思勤.声强直接测定法在发动机故障诊断中的应用[J].农业机械学报,2007,38(3):171-175.
[22]吕琛,王桂增,邱庆刚.基于声信号小波包分析的故障诊断[J].自动化学报,2004,30(4):554-559.
[23] Abe T,Anderton D.Digital acoustic intensity techniques in gasoline engine nosise studies [C].SAE Paper 820363,1982.
[24] Takaaki Musha,Jurric Taniguchi.Measurement of sound intensity using a single moving microphone [J].Appliced Acoustics,2005,66(5):554-559.
[25]熊敏.基于发动机缸盖声压的综合诊断系统[D].北京科技大学硕士学位论文,2004,8.
[26]余成渊.基于缸盖振动和缸压的发动机诊断的研究[D].北京科技大学硕士学位论文,2004,6.
[27]刘永志,王建中.摩托车发动机常见异响的判断及预防[J].摩托车,2005:33-36.
[28]阎平凡.人工神经网络与模拟进行计算[M].北京:清华大学出版社,2002
[29] Martin T.Hagan,Howard B.Demuth,Mark Beale.Neural Network Design[M].Beijing:Chian Machine Press,2002.
[30]沈世缢.神经网络系统理论及其应用[M].北京:科学出版社,1998.
[31]何春香.小波与BP神经网络在零件识别应用中的研究[D].哈尔滨理工大学硕士学位论文,2008,3.
[32]张立明.人工神经网络的模型及其应用[M].上海:上海复旦大学出版社,2003.
[33]邱忠宇等.基于知识和模糊神经网络的故障诊断技术[J].热能动力工程,1999,14(9):390-392.
[34] Jian Guan,James H and Graham. An integrated approach for fault diagnosis with learning[J] Computers in Industry.1996,32(1):33-51.
[35]毕美华等.基于模糊神数学和RBF神经网络的事故隐层[J].热能动力工程.2007,15(7).
[36]黄巧巧.基于BP神经网络的手写数字识别系统研究[D].华中师范大学硕士论文,2009.
[37]潘国荣,谷川.BP算法改进及其在变形数据处理中的应用[J].同济大学学报(自然科学版)2006(1):118-121.
[38]林国玺,宣慧玉.遗传算法和BP人工神经网络在税收预测中的应用[J].系统工程理论方法应用,2005(4):145.
[39]陈桦,程云艳.BP神经网络算法的改进及在MATLAB中的实现[J].陕西科技大学学报,2004(4):45.
[40]张克正.基于遗传神经网络和光谱分析的船舶机械状态监测研究[D].大连海事大学硕士论文,2010.
[41]马增辉.火电厂燃煤管理系统智能控制的研究[D].华北电力大学硕士论文,2007.
[42]陈文.基于神经网络的航空发动机间隙对EGT影响的研究[D].中国民航大学硕士论文,2004.
[43]刘秀芝.基于模式识别理论的电机故障诊断技术研究[D].山东科技大学硕士论文,2005.
[44]齐校含.MATLAB平台上的基于车辆声信号的交通量在线统计原理性系统[D].东北师范大学硕士论文,2008.
[45]于婷婷.基于BP神经网络的滚动轴承故障诊断方法[D].大连理工大学硕士论文,2008.
[46] ERRY S W,WYBER R J.A Hopfield neural network approach for the reconstruction of wide-bandwidth sonar data[J].Neural Networks for Signal processing,2000,2:876-885.
[47]燕碧娟.振动筛损伤的声发射信号处理方法研究[D].太原理工大学硕士论文,2006
[48]王巧花,叶平,黄民.基于MATLAB的图形用户界面(GUI)设计[J].煤矿机械,2005(03):60-61.
[49]姚秀芳,崔松菲.基于MATLAB GUIDE的程序设计[J].电脑知识与技术,2009,5(27).
[50]张智星.MATLAB程序设计与应用[M].北京:清华大学出版社,2002.
[2]张德丰等.MATLAB神经网络应用设计[M].机械工业出版社,2009.
[3]任志英.基于声信号技术的发动机故障诊断系统研究[D].福州大学硕士学位论文,2005.
[4] D. W. Thomas, B. R. Wilkins: The analysis of vehicle sounds for recognition, Pattern Recognition[J].1972.l4, 379-389.
[5] J.C. Webster, M. M. Woodhead, A. Carpenter: Identifying diesel engine sounds [J]. Journal of Sound Vibration,Vo1.9, No.2,2 41-246,1969.
[6] Richard H.L yon. Machinery Noise and Diagnostics [M].LONDON:Buterworth,1987.
[7] S.K.Lee, P.R.White. Higher-order and time-frequency analysis and its application to fault detection in rotating machinery[J].Mechanical Systems and Signal Processing,1997, 11,(4):637-650.
[8]孙宗新.基于混沌理论的发动机异响信号提取研究[D].吉林大学硕士学位论文,2007.6.
[9]舒大文等.用振动和噪声信号对汽车变速箱齿轮的研究[J].昆明理工大学学报,1997,4.
[10]李建秋,欧阳明高.汽车发动机瞬时转速测量中的噪声分析及补偿方法[J].中国机械工程,2000,7.
[11]刘福.基于声音的发动机诊断方法研究与实践[D].东北大学硕士学位论文,2006,6.
[12] Lin Jing,Qu Liang-sheng,Feature extraction based on Morlet wavelet and its application for mechanical fault diagnosis[J].Journal of Vibration,2000.234(1):135-148.
[13] Liu Baoling,Shih-fu,Meng Qingfeng,Machinery diagnosis based on wavelet packets[J].Journal of Vibration and Control,1997,3(1):5-7.
[14] Yen,Gary G,Lin Kuo-Chung,Conditional health monitoring using vibration signatures[C].Proceedings of the IEEE Conference on Decision and Control,1999,5:4493-4498.
[15] Yen,Gary G,Lin Kuo-Chung,Wavelet packet feature extraction for vibration monitoring[C].IEEE Transactions on Industrial Electronics,2000,47(3):650-667
[16] Hambaba A,Huff E,Multiresolution error detection on early fatigur cracks in gears[C].IEEE Aerospace Conference Proceeding,2000,6:367-373
[17]赵纪元,何正嘉,孟庆丰等.小波包—自回归谱分析在振动诊断中的应用[J].振动工作学报,1995,8(3):198-203.
[18]葛郢汉.发动机异响的分析与诊断[J].汽车博士,2007,(2):37-38.
[19]周志新.摩托车顶杆式发动机异响分析[J].摩托车,2003:49-50.
[20]刘永志,王建中.摩托车发动机常见异响的判断及预防[J].摩托车,2005:33-36.
[21]魏少华,陈效华,常思勤.声强直接测定法在发动机故障诊断中的应用[J].农业机械学报,2007,38(3):171-175.
[22]吕琛,王桂增,邱庆刚.基于声信号小波包分析的故障诊断[J].自动化学报,2004,30(4):554-559.
[23] Abe T,Anderton D.Digital acoustic intensity techniques in gasoline engine nosise studies [C].SAE Paper 820363,1982.
[24] Takaaki Musha,Jurric Taniguchi.Measurement of sound intensity using a single moving microphone [J].Appliced Acoustics,2005,66(5):554-559.
[25]熊敏.基于发动机缸盖声压的综合诊断系统[D].北京科技大学硕士学位论文,2004,8.
[26]余成渊.基于缸盖振动和缸压的发动机诊断的研究[D].北京科技大学硕士学位论文,2004,6.
[27]刘永志,王建中.摩托车发动机常见异响的判断及预防[J].摩托车,2005:33-36.
[28]阎平凡.人工神经网络与模拟进行计算[M].北京:清华大学出版社,2002
[29] Martin T.Hagan,Howard B.Demuth,Mark Beale.Neural Network Design[M].Beijing:Chian Machine Press,2002.
[30]沈世缢.神经网络系统理论及其应用[M].北京:科学出版社,1998.
[31]何春香.小波与BP神经网络在零件识别应用中的研究[D].哈尔滨理工大学硕士学位论文,2008,3.
[32]张立明.人工神经网络的模型及其应用[M].上海:上海复旦大学出版社,2003.
[33]邱忠宇等.基于知识和模糊神经网络的故障诊断技术[J].热能动力工程,1999,14(9):390-392.
[34] Jian Guan,James H and Graham. An integrated approach for fault diagnosis with learning[J] Computers in Industry.1996,32(1):33-51.
[35]毕美华等.基于模糊神数学和RBF神经网络的事故隐层[J].热能动力工程.2007,15(7).
[36]黄巧巧.基于BP神经网络的手写数字识别系统研究[D].华中师范大学硕士论文,2009.
[37]潘国荣,谷川.BP算法改进及其在变形数据处理中的应用[J].同济大学学报(自然科学版)2006(1):118-121.
[38]林国玺,宣慧玉.遗传算法和BP人工神经网络在税收预测中的应用[J].系统工程理论方法应用,2005(4):145.
[39]陈桦,程云艳.BP神经网络算法的改进及在MATLAB中的实现[J].陕西科技大学学报,2004(4):45.
[40]张克正.基于遗传神经网络和光谱分析的船舶机械状态监测研究[D].大连海事大学硕士论文,2010.
[41]马增辉.火电厂燃煤管理系统智能控制的研究[D].华北电力大学硕士论文,2007.
[42]陈文.基于神经网络的航空发动机间隙对EGT影响的研究[D].中国民航大学硕士论文,2004.
[43]刘秀芝.基于模式识别理论的电机故障诊断技术研究[D].山东科技大学硕士论文,2005.
[44]齐校含.MATLAB平台上的基于车辆声信号的交通量在线统计原理性系统[D].东北师范大学硕士论文,2008.
[45]于婷婷.基于BP神经网络的滚动轴承故障诊断方法[D].大连理工大学硕士论文,2008.
[46] ERRY S W,WYBER R J.A Hopfield neural network approach for the reconstruction of wide-bandwidth sonar data[J].Neural Networks for Signal processing,2000,2:876-885.
[47]燕碧娟.振动筛损伤的声发射信号处理方法研究[D].太原理工大学硕士论文,2006
[48]王巧花,叶平,黄民.基于MATLAB的图形用户界面(GUI)设计[J].煤矿机械,2005(03):60-61.
[49]姚秀芳,崔松菲.基于MATLAB GUIDE的程序设计[J].电脑知识与技术,2009,5(27).
[50]张智星.MATLAB程序设计与应用[M].北京:清华大学出版社,2002.