提出了一种结合卷积神经网络，小波变换和奇异值分解理论的水电机组故障诊断方法。利用卷积神经网络提取机组轴心轨迹的图像特征；通过离散小波变换对摆度信号进行分解，获得信号的小波分解系数，对各分支系数进行重构，构造奇异值分解输入矩阵，提取矩阵奇异值作为特征向量。将两种方法提取的特征进行组合，构建包含图像特征和波形特征的混合特征向量，通过概率神经网络进行识别分类。为验证该方法的有效性，将水电机组常见故障在转子试验台上进行模拟，用上述方法进行诊断。结果表明，文中所提出的故障诊断方法能很好地识别水电机组不同运行状态，可为水电机组的故障诊断提供有效依据。
Abstract
A novel method of fault diagnosis based on Convolutional Neural Network (CNN), Discrete Wavelet Transform (DWT) and Singular Value Decomposition (SVD) is proposed in this paper. CNN is used to extract features of shaft orbit images, DWT is used to transform the de-noised swing signal of rotating machinery and the wavelet decomposition coefficients of each branch of the signal are obtained by the transformation. The SVD input matrix is formed after single branch reconstructing of the different branch coefficients, and the singular value is extracted to obtain the feature vector. The features extracted from both methods are combined and then classified by Probabilistic Neural Network. The results show that this hybrid method can identify different operating states of hydropower units and provide an effective basis for fault diagnosis.
关键词
水电机组 /
轴心轨迹 /
卷积神经网络 /
小波变换 /
奇异值分解 /
故障诊断
{{custom_keyword}} /
Key words
hydropower unit /
shaft orbit /
convolutional neural network /
wavelet transform /
singular value decomposition /
fault diagnosis
{{custom_keyword}} /
基金
复杂多变网构下水电机组稳定性机理与机网协调控制研究;基于多小波与贝叶斯网络的水电机组故障诊断研究;国家电网公司科技项目“技术中心抽蓄机组状态大数据平台规范及数据挖掘研究与应用”
{{custom_fund}}
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1]刘东,王昕,黄建荧,等.基于贝叶斯网络的水电机组振动故障诊断研究[J].水力发电学报,2019,38(2):112-120.
[2]周福成,唐贵基,何玉灵.基于改进VMD的风电齿轮箱不平衡故障特征提取[J].振动与冲击,2020,39(5):170-176.
[3]陈强强,戴邵武,戴洪德,等.基于IMF特征提取的滚动轴承故障诊断[J].噪声与振动控制,2020,40(1):46-50.
[4]杨宇,于德介,程军圣.基于EMD的奇异值分解技术在滚动轴承故障诊断中的应用[J].振动与冲击,2005(2):12-15,145.
[5]刘东,王昕,黄建荧,等.基于SVD的水电机组振动信号去噪方法研究[J].水电与新能源,2018,32(7):38-43.
[6]魏东,龚庆武,来文青,等.基于卷积神经网络的输电线路区内外故障判断及故障选相方法研究[J].中国电机工程学报,2016,36(S1):21-28.
[7]JEONG H, PARK S, WOO S. et al. Rotating machinery diagnostics using deep learning on orbit plot images[C]∥ 44th North American Manufacturing Research Conference (NAMRC), 2016,5:1 107-1 118.
[8]张达峰. 基于深度卷积神经网络的文字识别算法研究[D].贵阳：贵州大学, 2019.
[9]ABDEL-HAMID O, MOHAMED AR; JIANG H. et al. Convolutional neural networks for speech recognition [J]. IEEE-ACM Transactions on Audio Speech and Language Processing, 2014,22(10):1 533-1 545.
[10]胡茑庆,陈徽鹏,程哲,等.基于经验模态分解和深度卷积神经网络的行星齿轮箱故障诊断方法[J].机械工程学报,2019,55(7):9-18.
[11]K Phim. Deep learning for beginners[M]. New York: Springer Science+Business Media, 2017:124.
[12]刘东,王昕,黄建荧,等.基于小波变换与SVD的水电机组振动信号特征提取研究[J].中国农村水利水电,2018(12):169-172.
[13]MALLAT S. A theory for multi-resolution approximation: The wavelet approximation[J]. IEEE Trans, 1989(11):674-693.
[14]MALLAT S. A wavelet tour of signal processing[M]. Pittsburgh: Academic Press, 1998:255.
PDF(5233 KB)
鄂公网安备 42010602001274号 鄂ICP备14015464号-1