img

QQ群聊

img

官方微信

高级检索

黄金科学技术 ›› 2022, Vol. 30 ›› Issue (2): 209-221.doi: 10.11872/j.issn.1005-2518.2022.02.162

• 采选技术与矿山管理 • 上一篇    下一篇

基于改进CEEMDAN-DCNN的声发射源识别分类方法

谢学斌(),刘涛(),张欢   

  1. 中南大学资源与安全工程学院,湖南 长沙 410083
  • 收稿日期:2021-11-03 修回日期:2021-12-28 出版日期:2022-04-30 发布日期:2022-06-17
  • 通讯作者: 刘涛 E-mail:xbxie@csu.edu.cn;849130941@qq.com
  • 作者简介:谢学斌(1968-),男,湖南祁东人,教授,从事矿山地压与岩爆灾害的预测和控制技术研究工作。xbxie@csu.edu.cn
  • 基金资助:
    广西重点研发计划项目“地下矿山大型复杂采空区群灾害性地压智能监控预警与控制技术研究”(编号:桂科AB18294004)资助

Identification and Classification Method of Underground AE Source Based on Improved CEEMDAN-DCNN

Xuebin XIE(),Tao LIU(),Huan ZHANG   

  1. School of Resources and Safety Engineering,Central South University,Changsha 410083,Hunan,China
  • Received:2021-11-03 Revised:2021-12-28 Online:2022-04-30 Published:2022-06-17
  • Contact: Tao LIU E-mail:xbxie@csu.edu.cn;849130941@qq.com

摘要:

声发射源的准确分类识别是声发射地压监测预报预警研究的重要基础。针对矿山井下围岩体声发射事件信号和采掘作业噪声信号分类识别问题,提出了一种基于改进完备总体经验模态分解和深度卷积神经网络(DCNN)的智能识别分类方法。首先,对信号进行改进CEEMDAN降噪处理,即利用相关性系数阈值和排列熵(PE)阈值剔除伪分量和噪声分量;然后,利用DCNN对降噪后的信号自动提取高维特征;最后,将特征用于softmax分类器分类识别,实现智能化井下信号源多分类。研究表明:改进CEEMDAN能够有效剔除伪分量及噪声分量;相比其他机器学习方法,改进CEEMDAN-DCNN方法具有准确率高和稳定性较好等优点。信号源识别分类方法研究为地压监测预警预报提供了重要的基础数据,准确的灾害预警预报可为矿山井下作业人员和设备提供安全保障。

关键词: 声发射监测, 波形分类, 信号分类识别, 改进CEEMDAN, 深度卷积神经网络(DCNN), 排列熵(PE)

Abstract:

Accurate classification and identification of acoustic emission sources is an important basis for the study of acoustic emission ground pressure monitoring, forecasting and early warning.Aiming at the clas-sification and identification of acoustic emission event signals and mining operation noise signals of surrounding rock masses in underground mines, an intelligent recognition and classification method based on improved complete ensemble empirical mode decomposition and deep convolutional neural network(DCNN)was proposed.Firstly,the signal was decomposed by CEEMDAN, the decomposed IMF components were screened, and the components greater than the permutation entropy threshold or less than the correlation coefficient threshold were removed, and the residual IMF components were reconstructed to obtain the denoised waveform.Then, the DCNN method was used to automatically extract high-dimensional features from the denoised waveform.Finally, the features were used for classification and recognition of softmax classifier to realize intelligent multi-classification of underground signal sources.The results of this research show that:(1)Aiming at the difficulty of multi-classification of waveforms received by acoustic emission monitoring equipment,a waveform classification and recognition method based on improved CEEMDAN-DCNN is proposed.Combined with the advantages of improved CEEMDAN’s advantages of adaptive analysis,pro-cessing of nonlinear and non-stationary signals and the ability of DCNN to automatically extract high-dimensional features, the intelligent multi-classification of underground signal sources is realized.(2)In order to verify the advantages of the improved CEEMDAN algorithm, the simulation signal is constructed to simulate the acoustic emission signal of surrounding rock mass containing noise signal, and the background noise component and pseudo component are eliminated by a joint threshold.The results show that the improved CEEMDAN algorithm can eliminate noise signals and some false components, and retain the essential characteristics of the signal.(3)Through the test, the accuracy of waveform classification based on the improved CEEMDAN-DCNN method in this paper reaches 97.12%. Compared with the traditional SVM, ANN, and CNN methods, the accuracy of waveform classification is higher and the stability is better. The accuracy of DCNN classification and recognition is improved dueing to the signal preprocessed by improved CEEMDAN.(4)The waveform recognition and classification method in this paper can accurately identify the acoustic emission events of surrounding rock masses and non-surrounding rock masses, provide reliable basic research data for ground pressure monitoring and early warning models, and increase the accuracy of ground pressure monitoring and safety early warning and forecasting.

Key words: acoustic emission monitoring, waveform classification, signal classification and recognition, improved CEEMDAN, deep convolutional neural network(DCNN), permutation entropy(PE)

中图分类号: 

  • TD76

图1

改进CEEMDAN-DCNN波形识别分类模型"

表1

CEEMD、CEEMDAN及改进CEEMDAN法参数及性能比较"

方法噪声个数噪声幅值ai噪声标准差迭代次数嵌入维数时间延迟计算耗时/s正交性指标
CEEMD35×20.2----23.07980.0879
CEEMDAN700.20.1200--8.53360.0318
改进CEEMDAN700.20.1200614.00673.8396e-4

图2

仿真信号时域波形"

图3

仿真信号CEEMD分解结果"

图4

仿真信号CEEMDAN分解结果"

图5

仿真信号改进CEEMDAN分解结果"

图6

声发射监测系统网络结构示意图"

图7

4类典型信号波形图"

表2

改进CEEMDAN分解参数"

方法

噪声

个数

噪声

幅值ai

噪声

标准差

迭代

次数

嵌入

维数

时间

延迟

改进CEEMDAN700.20.120061

图8

围岩体声发射波形CEEMDAN分解结果"

表3

IMF分量的PE值和相关系数"

分量排列熵值相关性系数
IMF10.910.14
IMF20.740.92
IMF30.680.7
IMF40.670.61
IMF50.580.58
IMF60.500.38
IMF70.420.30
IMF80.210.15

图9

围岩体声发射信号改进CEEMDAN分解结果"

图10

围岩体声发射事件原始波形及降噪后信号对比"

表4

DCNN模型主要结构参数"

网络层输出卷积核尺寸/步长padding激活函数
输入层1×1 024---
卷积层C11×641×3/1sameReLU
DropoutD11×64比率:0.2
卷积层C21×2561×3/1sameReLU
DropoutD21×256比率:0.2
卷积层C31×321×3/1sameReLU
DropoutD31×32比率:0.2
卷积层C41×321×3/1sameReLU
DropoutD41×32比率:0.2
卷积层C51×321×3/1sameReLU
DropoutD51×32比率:0.2
卷积层C61×321×3/1sameReLU
DropoutD61×32比率:0.2
全连接层F11×512--ReLU
dropout1×512比率:0.5
全连接层F21×4---

表5

本文方法与SVM、ANN及CNN性能对比"

评估指标本文方法DCNNSVMANNCNN
准确率/%97.12(σ=0.89%93.44(σ=1.02%84.11(σ=1.99%69.77(σ=3.22%89.23(σ=1.87%
精确率/%97.25(σ=0.71%93.35(σ=1.16%84.09(σ=1.96%68.26(σ=3.19%89.13(σ=2.03%
召回率/%98.09(σ=0.83%91.43(σ=1.19%83.52(σ=1.98%68.83(σ=3.18%90.69(σ=1.91%
计算时间/s5.67(σ=0.52%7.72(σ=0.52%20.19(σ=0.52%22.10(σ=0.52%10.65(σ=0.52%

图11

测试集分类准确率和损失率函数曲线"

Albert A, Nii A,2010. A criterion of selecting relevant intrinsic mode functions in empirical mode decomposition[J].Advances in Adaptive Data Analysis,2(1):1-24.
Chen Bingrui, Wu Hao, Chi Xiuwen,et al,2019. Real-time recognition algorithm for microseismic signals of rock failure based on STA/LTA and its engineering application[J].Rock and Soil Mechanics,40(9):3689-3696.
Cheng Tiedong, Wu Yiwen, Luo Xiaoyan,et al,2019. Method for feature extraction and classification of mine microseismic signals based on EWT_Hankel_SVD[J].Chinese Journal of Scientific Instrument,40(6):181-191.
Dong Longjun, Sun Daoyuan, Li Xibing,et al,2016.A statistical method to identify blasts and microseismic events and its engineering application[J].Chinese Journal of Rock Me-chanics and Engineering,35(7):1423-1433.
Dong Longjun, Tang Zheng, Li Xibing,et al,2020.Discrimination of mining microseismic events and blasts using convolutional neural networks and original waveform[J].Journal of Central South University,27(10):3078-3089.
Gao Junyu, Yang Xiaoshan, Zhang Tianzhu,et al,2016.Robust vision tracking method via deep learning[J].Chinese Journal of Computers,39(7):1419-1434.
Hao Yongmei, Du Zhanghao, Yang Wenbin,et al,2019. Pipeline leakage signal recognition based on improved ELMD and multi-scale entropy[J].Chinese Journal of Safety Scie-nce,29(8):105-111.
He Li, Zheng Zaoxian, Xiang Fengtao,et al,2021. Advances in text classification technology based on deep learning[J]. Computer Engineering,47(2):1-11.
Hu Jianqing, Chen Huipeng, Cheng Zhe,et al,2019.Fault diagnosis for planetary gearbox based on EMD and deep convolutional neural networks[J].Journal of Mechanical Engin-eering,55(7):9-18.
Jiang Wenwu, Yang Zuolin, Xie Jianmin,et al,2015. Application of FFT spectrum analysis to identify microseismic signals[J].Science and Technology Review,33(2):86-90.
Kortstrom J, Uski M, Tiira T,2016.Automatic classification of seismic events within a regional seismograph network[J].Computers and Geosciences,87:22-30.
Li Hong, Liu Fang, Yang Shuyuan,et al,2016.Remote sensing image fusion based on deep support value learning networks[J].Chinese Journal of Computers,39(8):1583-1596.
Li Ting,2017.Research in Locomotive Bearing Fault Diagnosis Method Based on Signal Modal Decomposition[D].Xi’an:Chang’an University.
Li Wei,2017.Feature extraction and classification method of mine microseismic signals based on LMD and pattern recognition[J]. Journal of China Coal Society,42(5):1156-1164.
Liao Zhiqin, Wang Liguan, He Zhengxiang,2020.Feature extraction and classification of mine microseismic signals based on EEMD and correlation dimension[J]. Gold Science and Technology,28(4):585-594.
Liu Jianpo, Li Yuanhui, Zhang Fengpeng,et al,2013. Stability analysis of rockmass based on acoustic emission monitoring in deep stope[J].Journal of Mining and Safety Engineering,30(2):243-250.
Peng P A, He Z X, Wang L G,et al,2020.Automatic classification of microseismic records in underground mining:A deep learning approach[J]. IEEE Access,8:17863-17876.
Peng Yunsai, Xia Fei, Yuan Bo,et al,2020. Fault diagnosis of traction battery pack based on improved convolution neural network and information fusion[J].Automotive Engineering,42(11):1529-1535.
Wu Z H, Huang N E,2009.Ensemble empirical mode decomposition:A noise assisted data analysis method[J].Advances in Adaptive Data Analysis,1:1-41.
Zhao Guoyan, Deng Qinglin, Li Xibing,et al,2017. Recognition of microseismic waveforms based on EMD and morphological fractal dimension[J]. Journal of Central South University(Science and Technology),48(1):162-167.
Zheng Jinde, Cheng Junsheng, Yang Yu,2013.Modified EEMD algorithm and its applications[J].Journal of Vibration and Shock,32(21):21-26.
Zhou Feiyan, Jin Linpeng, Dong Jun,2017. A review of convolutional neural networks[J].Chinese Journal of Computers,40(6):1229-1251.
Zhu Quanjie, Jiang Fuxing, Yu Zhengxing,et al,2012. Study on energy distribution characters about blasting vibration and rock fracture microseismic signal[J].Chinese Journal of Rock Mechanics and Engineering,31(4):723-730.
陈炳瑞,吴昊,池秀文,等,2019.基于STA/LTA岩石破裂微震信号实时识别算法及工程应用[J].岩土力学,40(9):3689-3696.
程铁栋,吴义文,罗小燕,等,2019.基于EWT_Hankel_SVD的矿山微震信号特征提取及分类方法[J].仪表仪器学报,40(6):181-191.
董陇军,孙道元,李夕兵,等,2016.微震与爆破事件统计识别方法及工程应用[J].岩石力学与工程学报,35(7):1423-1433.
董陇军,唐正,李夕兵,等,2020.基于卷积神经网络与原始波形的微震与爆破事件辨识方法[J].中南大学学报,27(10):3078-3089.
高君宇,杨小汕,张天柱,等,2016.基于深度学习的鲁棒性视觉跟踪方法[J].计算机学报,39(7):1419-1434.
郝永梅,杜璋昊,杨文斌,等,2019.基于改进ELMD和多尺度熵的管道泄漏信号识别[J].中国安全科学学报,29(8):105-111.
何力,郑灶贤,项凤涛,等,2021.基于深度学习的文本分类技术研究进展[J].计算机工程,47(2):1-11.
胡茑庆,陈徽鹏,程哲,等,2019.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,55(7):9-18.
江文武,杨作林,谢建敏,等,2015.FFT频谱分析在微震信号识别中的应用[J].科技导报,33(2):86-90.
李红,刘芳,杨淑媛,等,2016.基于深度支撑值学习网络的遥感图像融合[J]. 计算机学报,39(8):1583-1596.
李婷,2017.基于信号模态分解的机车轴承故障诊断方法研究[D].西安:长安大学.
李伟,2017.基于LMD和模式识别的矿山微震信号特征提取及分类方法[J].煤炭学报,42(5):1156-1164.
廖智勤,王李管,何正祥,2020.基于EEMD和关联维数的矿山微震信号特征提取和分类[J].黄金科学技术,28(4):585-594.
刘建坡,李元辉,张凤鹏,等,2013.基于声发射监测的深部采场岩体稳定性分析[J].采矿与安全工程学报,30(2):243-250.
彭运赛,夏飞,袁博,等,2020.基于改进CNN和信息融合的动力电池组故障诊断方法[J]. 汽车工程,42(11):1529-1535.
赵国彦,邓青林,李夕兵,等,2017. 基于EMD和形态分形维数的微震波形识别[J]. 中南大学学报(自然科学版),48(1):162-167.
郑近德,程军圣,杨宇,2013.改进的 EEMD 算法及其应用研究[J].振动与冲击,32(21):21-26.
周飞燕,金林鹏,董军,2017.卷积神经网络研究综述[J].计算机学报,40(6):1229-1251.
朱权洁,姜福兴,于正兴,等,2012.爆破震动与岩石破裂微震信号能量分布特征研究[J].岩石力学与工程学报,31(4):723-730.
[1] 邵良杉,闻爽爽. 基于GRU神经网络的巷道平均风速获取研究[J]. 黄金科学技术, 2021, 29(5): 709-718.
[2] 毕林,周超,姚鑫. 基于视频序列的矿卡司机不安全行为识别[J]. 黄金科学技术, 2021, 29(1): 14-24.
[3] 王牧帆,罗周全,于琦. 基于 Stacking 模型的采空区稳定性预测[J]. 黄金科学技术, 2020, 28(6): 894-901.
[4] 廖智勤, 王李管, 何正祥. 基于EEMD和关联维数的矿山微震信号特征提取和分类[J]. 黄金科学技术, 2020, 28(4): 585-594.
[5] 党明智,张君,贾明涛. 黄土坡铜锌矿微震监测技术应用与灾害预警方法研究[J]. 黄金科学技术, 2020, 28(2): 246-254.
[6] 随晓丹,罗周全,秦亚光,王玉乐,彭东. 基于小波分解的尾矿坝浸润线预测方法研究[J]. 黄金科学技术, 2019, 27(1): 137-143.
[7] 张二洋,陈建宏. 基于Surpac矿山设计软件及虚幻引擎实现的矿山虚拟现实漫游系统[J]. 黄金科学技术, 2017, 25(4): 93-98.
[8] 刘晓明,赵君杰,彭平安,毕林,代碧波. 有效微震信号自动识别技术研究[J]. 黄金科学技术, 2017, 25(3): 84-91.
[9] 聂兴信,张国丹. 基于熵值法—突变理论的地下矿山紧急避险系统可靠性研究[J]. 黄金科学技术, 2016, 24(6): 72-77.
[10] 王婷玉,罗周全,秦亚光,孙杨. 主溜井垮塌三维探测及可视化分析与计算[J]. 黄金科学技术, 2016, 24(1): 97-101.
[11] 尹土兵,王品,张鸣鲁. 基于AHP及模糊综合评判的地下金属矿山安全分析与评价[J]. 黄金科学技术, 2015, 23(3): 60-66.
[12] 秦亚光,罗周全,周吉明,汪伟,孙杨. 采空区可视化集成系统信息管理研究与应用[J]. 黄金科学技术, 2015, 23(2): 57-62.
[13] 赵永未,王京海,吕春堂,杨力纲,朱会凯. Dini03数字水准仪在望儿山采空区沉降监测及地质灾害预测分析中的应用[J]. 黄金科学技术, 2013, 21(6): 68-72.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!