基于VMD和CNN−BiLSTM的矿井提升电动机故障诊断方法

李敬兆; 何娜; 张金伟; 王擎; 李化顺

doi:10.13272/j.issn.1671-251x.2022120065

基于VMD和CNN−BiLSTM的矿井提升电动机故障诊断方法

doi: 10.13272/j.issn.1671-251x.2022120065

李敬兆^1,,
何娜^1, ,,
张金伟²,
王擎³,
李化顺¹

1.
安徽理工大学电气与信息工程学院, 安徽淮南　232000
2.
安徽理工大学计算机科学与工程学院, 安徽淮南　232000
3.
中国电子信息产业集团有限公司，广东深圳　518057

基金项目: 国家自然科学基金项目（51874010）；淮南市科技计划项目（2021A243）。

详细信息

作者简介:
李敬兆（1964—），男，安徽淮南人，教授，博士研究生导师，博士，主要研究方向为嵌入式系统、人工智能技术，E-mail:jzhli@aust.edu.cn

通讯作者:
何娜（1999—），女，安徽宿州人，硕士研究生，主要研究方向为信号处理、设备故障诊断，E-mail:hn200742@163.com

中图分类号: TD67
计量
- 文章访问数: 838
- HTML全文浏览量: 23
- PDF下载量: 36
- 被引次数: 0
出版历程
- 收稿日期: 2022-12-20
- 修回日期: 2023-07-20
- 网络出版日期: 2023-08-08

Fault diagnosis method for mine hoisting motor based on VMD and CNN-BiLSTM

1.
School of Electrical and Information Engineering, Anhui University of Science and Technology, Huainan 232000, China
2.
School of Computer Science and Engineering, Anhui University of Science and Technology, Huainan 232000, China
3.
China Electronics Corporation, Shenzhen 518057, China

摘要

摘要: 针对传统基于音频信号的电动机故障诊断方法获取电动机音频信号特征信息不足和故障诊断精度不高的问题，提出了一种基于优化的变分模态分解（VMD）和卷积神经网络CNN−双向长短期记忆（BiLSTM）的矿井提升电动机故障诊断方法。针对模态混叠和端点效应问题，采用鲸鱼算法（WOA）优化的VMD对电动机音频信号进行分解，将电动机音频信号分解为K个本征模态分量（IMF），经Pearson相关系数筛选后，提取主IMF分量的13维静态MFCC特征参数，为了获取信号的动态特征，提取13维静态MFCC的一阶差分和二阶差分系数，构成39维特征向量，从而把动静态特征结合起来，提高故障诊断性能。为了提高故障诊断精度，在CNN中引入BiLSTM层，CNN在空间维度上提取音频信号的局部特征，BiLSTM在时间维度上保留音频信号的双向时间序列信息，捕获音频信号长距离依赖关系，从而最大程度保留全局和局部特征。实验结果表明：① VMD分解的每个IMF分量都具有独立的中心频率且分布均匀，在频域上表现出稀疏性的特点，能够有效避免模态混叠问题；在IMF求解中，VMD分解通过镜像延拓的方式避免了经验模态分解（EMD）和集合经验模态分解（EEMD）中出现的端点效应问题。② 基于13维静态MFCC特征的故障诊断准确率为97.5%，基于39维动静态MFCC特征的故障诊断准确率比基于13维静态MFCC特征的故障诊断准确率提高了1.11%。③基于CNN−BiLSTM诊断模型的准确率达到98.61%，与目前通用诊断模型CNN，BiLSTM和CNN−LSTM相比，准确率分别提高5.83%，4.17%和3.89%。
- 矿井提升电动机故障诊断 /
- 音频信号 /
- 变分模态分解 /
- 梅尔倒谱系数 /
- CNN−BiLSTM
Abstract: The traditional motor fault diagnosis method based on the audio signal is insufficient to obtain the feature information of the motor audio signal and the fault diagnosis precision is not high. In order to solve the above problems, a mine motor fault diagnosis method based on optimized variational mode decomposition (VMD) and convolutional neural network CNN bidirectional long short-term memory (BiLSTM) is proposed. The whale algorithm (WOA) optimized VMD is used to decompose the motor audio signal to address the issues of modal aliasing and endpoint effects. The motor audio signal is decomposed into K intrinsic mode functions (IMF). After Pearson correlation coefficient screening, the 13-dimensional static MFCC feature parameters of the main IMF component are extracted. In order to obtain the dynamic features of the signal, the first and second-order difference coefficients of the 13-dimensional static MFCC are extracted to form a 39-dimensional feature vector. By combining dynamic and static features, the performance of fault diagnosis can be improved. In order to improve the precision of fault diagnosis, a BiLSTM layer is introduced into the CNN. The CNN extracts local features of the audio signal in the spatial dimension. The BiLSTM preserves bidirectional time series information of the audio signal in the temporal dimension. It captures long-distance dependencies of the audio signal, thereby maximizing the preservation of global and local features. The experimental results show the following points. ① Each IMF component of VMD decomposition has an independent center frequency and uniform distribution, and exhibits sparsity in the frequency domain. It can effectively avoid modal aliasing problems. In IMF solving, VMD decomposition avoids endpoint effects in empirical mode decomposition (EMD) and ensemble empirical mode decomposition (EEMD) through mirror extension. ② The fault diagnosis accuracy based on 13-dimensional static MFCC features is 97.5%. The fault diagnosis accuracy based on 39-dimensional dynamic and static MFCC features is 1.11% higher than that based on 13-dimensional static MFCC features. ③ The accuracy of the diagnostic model based on CNN-BiLSTM reaches 98.61%, which is 5.83%, 4.17%, and 3.89% higher than the current universal diagnostic models CNN, BiLSTM, and CNN-LSTM, respectively.
- fault diagnosis of mine hoist motor /
- audio signal /
- variational mode decomposition /
- Mel cepstrum coefficient /
- CNN-BiLSTM

HTML全文

图 1 WOA优化VMD参数流程

Figure 1. Flow of WOA-VMD

下载: 全尺寸图片幻灯片

图 2 MFCC特征提取流程

Figure 2. MFCC feature extraction process

下载: 全尺寸图片幻灯片

图 3 CNN−BiLSTM模型结构

Figure 3. Structure of CNN-BiLSTM Model

下载: 全尺寸图片幻灯片

图 4 采集装置

Figure 4. Acquisition device

下载: 全尺寸图片幻灯片

图 5 现场安装

Figure 5. Site installation

下载: 全尺寸图片幻灯片

图 6 滤波降噪前后音频信号波形

Figure 6. Waveform of audio signal before and after noise reduction

下载: 全尺寸图片幻灯片

图 7 VMD参数的WOA优化结果

Figure 7. WOA optimization results of VMD parameters

下载: 全尺寸图片幻灯片

图 8 VMD分解

Figure 8. VMD decomposition

下载: 全尺寸图片幻灯片

图 10 EEMD分解

Figure 10. EEMD decomposition

下载: 全尺寸图片幻灯片

图 9 EMD分解

Figure 9. EMD decomposition

下载: 全尺寸图片幻灯片

图 11 IMF相关系数值

Figure 11. Correlation coefficient of IMF

下载: 全尺寸图片幻灯片

图 12 不同模型损失函数曲线

Figure 12. Loss function curves of different models

下载: 全尺寸图片幻灯片

图 13 各模型诊断结果对比

Figure 13. Comparison of diagnostic results of each model

下载: 全尺寸图片幻灯片

图 14 测试集混淆矩阵

Figure 14. Test set confusion matrix

下载: 全尺寸图片幻灯片

表 1 各网络层参数

Table 1. Parameters of each network layer

网络层	主要参数
卷积层1	核大小：$ 5 \times 1 $，数量：32，步长：1
池化层1	核大小：$ 2 \times 1 $，步长：1
卷积层2	核大小：$ 3 \times 1 $，数量：64，步长：1
池化层2	核大小：$ 2 \times 1 $，步长：1
BiLSTM层	单元数：5
全连接层	—
Softmax层	—

下载: 导出CSV

表 2 最优VMD参数

Table 2. Optimal VMD parameters

数据类型	惩罚因子	模态个数
电流故障	2 064	5
正常状态	1 968	5
机械故障	2 532	6
超载故障	2 666	7

下载: 导出CSV

表 3 不同分解方式的相关系数

Table 3. Correlation coefficients of different decomposition modes

分解方式	IMF1	IMF2	IMF3	IMF4	IMF5	IMF6	IMF7	IMF8	IMF9
VMD	0.349 0	0.335 1	0.405 8	0.825 1	0.249 2	0.166 0	−	−	−
EMD	0.262 6	0.794 7	0.318 4	0.259 1	0.088 7	0.0387	−0.018 0	0.009 2	−
EEMD	0.195 4	0.686 4	0.421 3	0.201 1	0.194 2	0.214 0	0.054 2	0.032 2	0.022 5

下载: 导出CSV

表 4 不同模型评价结果

Table 4. Evaluation results of different models

模型类型	准确率/%	训练时间/s
CNN	92.78	9
BiLSTM	94.44	10
CNN−LSTM	94.72	12
CNN−BiLSTM	98.61	14

下载: 导出CSV

参考文献(21)

[1]	何俊峰,肖慧明. 矿井提升机健康管理系统研究[J]. 制造业自化化,2020,42(6):4-7,38. HE Junfeng,XIAO Huiming. Reaserch on system of maganement for mine hoist[J]. Manufacturing Automation,2020,42(6):4-7,38.
[2]	赵书涛,王二旭,陈秀新,等. 声振信号联合1D−CNN的大型电机故障诊断方法[J]. 哈尔滨工业大学学报,2020,52(9):116-122. ZHAO Shutao,WANG Erxu,CHEN Xiuxin,et al. Fault diagnosis method for large motor based on sound-vibration signal combined with 1D-CNN[J]. Journal of Harbin Institute of Technology,2020,52(9):116-122.
[3]	李伟,李硕. 理解数字声音——基于一般音频/环境声的计算机听觉综述[J]. 复旦学报(自然科学版),2019,58(3):269-313. LI Wei,LI Shuo. Understanding digital audio−a review of general audio/ambient sound based computer audition[J]. Journal of Fudan University(Natural Science),2019,58(3):269-313.
[4]	李海. 基于EMD和特征融合的电机故障诊断[D]. 杭州: 浙江大学, 2013. LI Hai. Faults diagnosis of motor based on EMD and feature-fusion[D]. Hangzhou: Zhejiang University, 2013.
[5]	孙杰臣. 基于音频的矿井提升机故障诊断和健康预测系统[D]. 淮南: 安徽理工大学, 2021. SUN Jiechen. Fault diagnosis and health prediction system for mine hoists based on audio signal[D]. Huainan: Anhui University of Science and Technology, 2021.
[6]	路敬祎,马雯萍,叶东,等. 基于VMD的音频信号增强算法研究[J]. 机械工程学报,2018,54(10):10-15. doi: 10.3901/JME.2018.10.010 LU Jingyi,MA Wenping,YE Dong,et al. Algorithm of sound signal enhancement based on VMD[J]. Journal of Mechanical Engineering,2018,54(10):10-15. doi: 10.3901/JME.2018.10.010
[7]	丁石川,厉雪衣,杭俊,等. 深度学习理论及其在电机故障诊断中的研究现状与展望[J]. 电力系统保护与控制,2020,48(8):172-187. DING Shichuan,LI Xueyi,HANG Jun,et al. Deep learning theory and its application to fault diagnosis of an electric machine[J]. Power System Protection and Control,2020,48(8):172-187.
[8]	马立玲,刘潇然,沈伟,等. 基于一种改进的一维卷积神经网络电机故障诊断方法[J]. 北京理工大学学报,2020,40(10):1088-1093. MA Liling,LIU Xiaoran,SHEN Wei,et al. Motor fault diagnosis method based on an improved one-dimensional convolutional neural network[J]. Transactions of Beijing Institute of Technology,2020,40(10):1088-1093.
[9]	张鹏,束小曼,厉雪衣,等. 基于LSTM的交流电机系统故障诊断方法研究[J]. 电机与控制学报,2022,26(3):109-116. ZHANG Peng,SHU Xiaoman,LI Xueyi,et al. LSTM-based fault diagnosis of AC electric machine system[J]. Electric Machines and Control,2022,26(3):109-116.
[10]	向玲,王朋鹤,李京蓄. 基于CNN−LSTM的风电机组异常状态检测[J]. 振动与冲击,2021,40(22):11-17. XIANG Ling,WANG Penghe,LI Jingxu. Abnormal state detection of wind turbines based on CNN-LSTM[J]. Journal of Vibration and Shock,2021,40(22):11-17.
[11]	李可,牛园园,宿磊,等. 参数优化VMD的滚动轴承故障诊断方法[J]. 振动工程学报,2023,36(1):280-287. LI Ke,NIU Yuanyuan,SU Lei,et al. Rolling bearing fault diagnosis method based on parameter optimized VMD[J]. Journal of Vibration Engineering,2023,36(1):280-287.
[12]	ZOSSO D,DRAGOMIRETSKIY K. Variational mode decomposition[J]. IEEE Transactions on Signal Processing:A Publication of the IEEE Signal Procession Society,2014,62(3):531-544.
[13]	MIRJALILI S,LEWIS A. The whale optimization algorithm[J]. Advances in Engineering Software,2016,95:51-67. doi: 10.1016/j.advengsoft.2016.01.008
[14]	曹仕骏,郑近德,潘海洋,等. 基于改进自适应经验傅里叶分解的滚动轴承故障诊断方法[J]. 振动与冲击,2022,41(15):287-299. CAO Shijun,ZHENG Jinde,PAN Haiyang,et al. Enhanced adaptive empirical Fourier decomposition based rolling bearing fault diagnosis method[J]. Journal of Vibration and Shock,2022,41(15):287-299.
[15]	王前,王刚,蒋晗晗,等. 基于MFCC与CDET的滚动轴承故障诊断方法研究[J]. 控制工程,2019,26(9):1682-1686. WANG Qian,WANG Gang,JIANG Hanhan et al. Study on fault diagnosis of rolling bearing based on MFCC and CDET[J]. Control Engineering of China,2019,26(9):1682-1686.
[16]	李宏全,郭兴明,郑伊能. 基于EMD和MFCC的舒张期心杂音的分类识别[J]. 振动与冲击,2017,36(11):8-13. LI Hongquan,GUO Xingming,ZHENG Yineng. Classification and recognition of diastolic heart murmurs based on EMD and MFCC[J]. Journal of Vibration and Shock,2017,36(11):8-13.
[17]	刘思思,谭建平,易子馗. 基于MFCC和SVM的车窗电机异常噪声辨识方法研究[J]. 振动与冲击,2017,36(5):102-107. LIU Sisi,TAN Jianping,YI Zikui. A window motor abnormal noiseidentification method based on MFCC and SVM[J]. Journal of Vibration and Shock,2017,36(5):102-107.
[18]	崔佳嘉,马宏忠. 基于改进MFCC和3D-CNN的变压器铁心松动故障声纹识别模型[J]. 电机与控制学报,2022,26(12):150-160. CUI Jiajia,MA Hongzhong. Voiceprint recognition model of transformer core looseness fault based on improved MFCC and 3D-CNN[J]. Electric Machines and Control,2022,26(12):150-160.
[19]	汪欣,毛东兴,李晓东. 基于声信号和一维卷积神经网络的电机故障诊断研究[J]. 噪声与振动控制,2021,41(2):125-129. WANG Xin,MAO Dongxing,LI Xiaodong. Motor fault diagnosis using microphones and one-dimensional convolutional neural network[J]. Noise and Vibration Control,2021,41(2):125-129.
[20]	董绍江,李洋,梁天,等. 基于CNN−BiLSTM的滚动轴承变工况故障诊断方法[J]. 振动. 测试与诊断,2022,42(5):1009-1016,1040. DONG Shaojiang,LI Yang,LIANG Tian,et al. Fault diagnosis method of rolling bearing based on CNN-BiLSTM under variable working conditions[J]. Journal of Vibration,Measurement & Diagnosis,2022,42(5):1009-1016,1040.
[21]	王宏伟,孙文磊,张小栋,等. 基于优化VMD复合多尺度散布熵及LSTM的风力发电机齿轮箱故障诊断方法研究[J]. 太阳能学报,2022,43(4):288-295. WANG Hongwei,SUN Wenlei,ZHANG Xiaodong,et al. Fault diagnosis method of wind turbine's gearbox based on composite multiscale dispersion entropy of optimised WMD and LSTM[J]. Acta Energiae Solaris Sinica,2022,43(4):288-295.