基于贝叶斯深度学习的外啮合齿轮泵健康状态识别研究

doi:10.3901/JME.2023.16.397

机械工程学报 ›› 2023, Vol. 59 ›› Issue (16): 397-405.doi: 10.3901/JME.2023.16.397

扫码分享

基于贝叶斯深度学习的外啮合齿轮泵健康状态识别研究

郭锐^1,2,3, 王涛¹, 李永涛¹, 王建伟^1,4, 蔡伟^1,3,4, 赵静一^1,3,4

1. 燕山大学河北省重型机械流体动力传输与控制重点实验室秦皇岛 066004;
2. 浙江大学流体动力与机电系统国家重点实验室杭州 310027;
3. 燕山大学先进锻压成形技术与科学教育部重点实验室秦皇岛 066004;
4. 河北省特种运载装备重点实验室秦皇岛 066004

收稿日期:2022-09-05 修回日期:2023-01-10 出版日期:2023-08-20 发布日期:2023-11-15
通讯作者: 郭锐(通信作者),男,1980年出生,教授。主要研究方向为流体动力基础件和机电装备电液控制系统的创新设计与可靠性研发。E-mail:guorui@ysu.edu.cn
基金资助:
国家自然科学基金(52075469,12173054);国家重点研发计划(2019YFB2005204);陕西省液压技术重点实验室开放基金(YYJS2022KF12)资助项目。

Health Status Recognition of Gear Pump Based on Bayesian-LSTM

GUO Rui^1,2,3, WANG Tao¹, LI Yongtao¹, WANG Jianwei^1,4, CAI Wei^1,3,4, ZHAO Jingyi^1,3,4

1. Hebei Key Laboratory of Heavy Machinery Fluid Power Transmission and Control, Yanshan University, Qinhuangdao 066004;
2. State Key Laboratory of Fluid Power and Electromechanical System, Zhejiang University, Hangzhou 310027;
3. Key Laboratory of Advanced Forging Technology and Science, Ministry of Education, Yanshan University, Qinhuangdao 066004;
4. Hebei Key Laboratory of Specialized Transportation Equipment, Qinhuangdao 066004

Received:2022-09-05 Revised:2023-01-10 Online:2023-08-20 Published:2023-11-15

摘要/Abstract

摘要： 液压泵健康状态的精准识别是液压元件预防性维护的一部分。采用贝叶斯算法对长短时记忆网络（Longshort-term memorg,LSTM）进行优化，构建一种基于贝叶斯长短时记忆网络（Bayesian-LSTM）的外啮合齿轮泵健康状态识别模型，完成了对外啮合齿轮泵健康状态的识别。首先运用改进的变分模态分解（Improved variational modal,IVMD）方法对蕴含丰富健康信息的样本泵振动信号进行分解重构，基于时域、频域及时频域对重构信号进行特征提取，将所选特征归一化处理后构造了能够表征振动信号的特征矩阵。最后将归纳好的特征矩阵与标签输入到Bayesian-LSTM（以下简称BDL）模型中进行训练，进而得到外啮合齿轮泵健康识别模型，为验证BDL模型在外啮合齿轮泵健康状态识别上的优越性，将其与LSTM模型及RNN模型进行对比。结果表明，BDL模型对样本泵的预测精度达到了98.3%以上，相比LSTM模型及RNN模型优势明显，可以为深度学习在液压元件状态识别的应用提供参考。

关键词: 外啮合齿轮泵, 健康状态识别, 加速退化试验, 贝叶斯长短时记忆网络

Abstract: Accurate recognition of the health status for hydraulic pumps is part of preventive maintenance（PM）. A health status recognition model of external gear pumps based on Bayesian-Long short-term memorg（BDL） combines the Bayesian method and the deep learning algorithm to realize the health state recognition of the external gear pumps is proposed. Firstly, improved variational modal decomposition（IVMD） is used to decompose and reconstruct the vibration signal of the sample pump, whose features are extracted based on the time domain, frequency domain and time-frequency domain. After the selected features are normalized, the feature matrix representing the vibration signal is constructed. Finally, the feature matrix and labels are used to train the BDL model,and the external gear pump health status recognition model is obtained. In order to verify the superiority of the BDL in recognition for the health status of external gear pumps, a comparison with models of LSTM and RNN has carried. The results show that the prediction accuracy of BDL is more than 98.3%, which has obvious advantages than LSTM and RNN. The conclusion can provide a reference for the application of deep learning in status recognition of hydraulic components.

Key words: external gear pump, health status recognition, accelerated degradation test, bayesian-LSTM

中图分类号:

TH325

郭锐, 王涛, 李永涛, 王建伟, 蔡伟, 赵静一. 基于贝叶斯深度学习的外啮合齿轮泵健康状态识别研究[J]. 机械工程学报, 2023, 59(16): 397-405.

GUO Rui, WANG Tao, LI Yongtao, WANG Jianwei, CAI Wei, ZHAO Jingyi. Health Status Recognition of Gear Pump Based on Bayesian-LSTM[J]. Journal of Mechanical Engineering, 2023, 59(16): 397-405.

参考文献

[1] 李壮云. 液压元件与系统[M]. 北京:机械工业出版社, 2011. LI Zhuangyun. Hydraulic components and systems[M]. Beijing:China Machine Press, 2011.
[2] KIM H E, HWANG S S, TAN A C C, et al. Integrated approach for diagnostics and prognostics of HP LNG pump based on health state probability estimation[J]. Journal of Mechanical Science & Technology, 2012, 26(11):3571-3585.
[3] ZHAO Rui, YAN Ruiqiang, WANG Jinjiang, et al. Learning to monitor machine health with convolutional Bi-directional LSTM networks[J]. Sensors, 2017, 17(2):273.
[4] LI Xiang, DING Qian, SUN Jianqiao. Remaining useful life estimation in prognostics using deep convolution neural networks[J]. Reliability Engineering and System Safety, 2018, 172:1-11.
[5] ZHAO Rui, YAN Ruiqiang, CHEN Zhenghua, et al. Deep learning and its applications to machine health monitoring[J]. Mechanical Systems and Signal Processing, 2019, 115:213-237.
[6] 吴丹,金敏. 基于容错度自适应支持向量机的液压泵故障诊断[J]. 中国机械工程, 2011, 22(21):2582-2587. WU Dan, JIN Min. Fault diagnosis of hydraulic pump based on fault tolerance adaptive support vector machine[J]. China Mechanical Engineering, 2011, 22(21):2582-2587.
[7] 何庆飞,陈桂明,陈小虎,等. 基于改进灰色神经网络的液压泵寿命预测[J]. 中国机械工程, 2013, 24(4):500-506. HE Qingfei, CHEN Guiming, CHEN Xiaohu, et al. Life prediction of hydraulic pump based on improved grey neural network[J]. China Mechanical Engineering, 2013, 24(4):500-506.
[8] VM A, VS B, VI C. Fault diagnosis of monoblock centrifugal pump using SVM[J]. Engineering Science and Technology, an International Journal, 2014, 17(3):152-157.
[9] GE W. Wear condition prediction of hydraulic pump[J]. Journal of Beijing University of Aeronautics & Astronautics, 2011, 37(11):1410-1414.
[10] 裴洪,胡昌华,司小胜,等. 基于机器学习的设备剩余寿命预测方法综述[J]. 机械工程学报, 2019, 55(8):1-13. PEI Hong, HU Changhua, SI Xiaosheng, et al. Review of residual life prediction method of equipment based on machine learning[J]. Journal of Mechanical Engineering, 2019, 55(8):1-13.
[11] BENGIO S, DENG L, LAROCHELLE H, et al. Guest editors' introduction:Special section on learning deep architectures[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(8):1795-1797.
[12] XIAO D, HUANG Y, WANG H, et al. Health assessment for piston pump using LSTM neural network[C]//2018 International Conference on Sensing, Diagnostics, Prognostics, and Control (SDPC). 2018, 131-137.
[13] BIE F, DU T, LYU F, et al. An integrated approach based on improved CEEMDAN and LSTM deep learning neural network for fault diagnosis of reciprocating pump[J]. IEEE Access, 2021, 9:23301-23310.
[14] 魏晓良,潮群,陶建峰,等. 基于LSTM和CNN的高速柱塞泵故障诊断[J]. 航空学报, 2021, 42(3):435-445. WEI Xiaoliang, CHAO Qun, TAO Jianfeng, et al. Cavition fault diagnosis method for highs-speed plunger pumps based on LSTM and CNN[J]. Actac Aeronautica et Astronautica Sinica, 2021, 42(3):435-445.
[15] BARALDI P, MANGILI F, ZIO E. Investigation of uncertainty treatment capability of model-based and data-driven prognostic methods using simulated data[J]. Reliability Engineering and System Safety, 2013, 112:94-108.
[16] SANKARARAMAN S. Significance, interpretation, and quantification of uncertainty in prognostics and remaining useful life prediction[J]. Mechanical Systems and Signal Processing, 2015, 52(53):228-247.
[17] STRAUBT D. Stochastic modeling of deterioration processes through dynamic bayesian networks[J]. Journal of Engineering Mechanics, 2009, 135(10):1089-1099.
[18] 罗亦泳,姚宜斌,黄城,等. 基于改进VMD的变形特征提取与分析[J]. 武汉大学学报(信息科学版), 2020, 45(4):612-619. LUO Yiyong, YAO Yibin, HUANG Cheng, et al. Deformation feature extraction and analysis based on improved VMD[J]. Journal of Wuhan University (Information Science Edition), 2020, 45(4):612-619.
[19] 张建伟,华薇薇,侯鸽. IVMD对泵站管道振动响应趋势的预测分析[J]. 振动测试与诊断, 2019, 39(3):478-483. ZHANG Jianwei, HUA Weiwei, HOU Ge. Prediction and analysis of ivmd on vibration response trend of pump station pipeline[J]. Journal of Vibration Measurement and Diagnosis, 2019, 39(3):478-483.
[20] BLEI D M, KUCUKELBIR A, MCAULIFFE J D. Variational Inference:A review for Statisticians[J]. Journal of the American Statistical Association, 2018, 112(518):859-877.

基于贝叶斯深度学习的外啮合齿轮泵健康状态识别研究

Health Status Recognition of Gear Pump Based on Bayesian-LSTM

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 7

编辑推荐

Metrics

本文评价

[1]	钟立强, 陈文华, 钱萍, 潘骏, 贺青川, 高亮. 基于连续应力与离散应力综合的加速退化试验设计方法研究[J]. 机械工程学报, 2020, 56(2): 201-209.
[2]	王浩伟, 周源, 滕飞, 盖炳良. 基于加速因子不变原则的加速退化试验优化设计[J]. 机械工程学报, 2018, 54(18): 212-219.
[3]	彭卫文;黄洪钟;李彦锋;杨园鉴;李海庆. 基于数据融合的加工中心功能铣头贝叶斯可靠性评估[J]. , 2014, 50(6): 185-191.
[4]	肖坤;顾晓辉;彭琛. 基于恒定应力加速退化试验的某引信用O型橡胶密封圈可靠性评估[J]. , 2014, 50(16): 62-69.
[5]	潘骏;王小云;陈文华;许仕伟;高亮;钱萍. 基于多元性能参数的加速退化试验方案优化设计研究[J]. , 2012, 48(2): 30-35.
[6]	陈循;张春华. 加速试验技术的研究、应用与发展[J]. , 2009, 45(8): 130-136.
[7]	汪亚顺;张春华;陈循;莫永强. 仿真基混合效应模型加速退化试验方案优化设计研究[J]. , 2009, 45(12): 108-114.