基于IKF-ARIMA-NARNN的滚动轴承噪声环境下寿命预测方法

李军星; 樊嘉辉; 王治华; 傅惠民

doi:10.13224/j.cnki.jasp.20230701

基于IKF-ARIMA-NARNN的滚动轴承噪声环境下寿命预测方法

doi: 10.13224/j.cnki.jasp.20230701

李军星^{1, 2, 3,},
樊嘉辉^1,,
王治华⁴,
傅惠民⁴

1.
河南科技大学机电工程学院，河南洛阳 471000
2.
河南科技大学机械装备先进制造河南省协同创新中心，河南洛阳 471000
3.
河南科技大学高端轴承河南省协同创新中心，河南洛阳 471000
4.
北京航空航天大学小样本技术研究中心，北京 100191

基金项目: 国家自然科学基金（52005159）；河南省科技研发计划联合基金青年科学家项目（225200810073）；河南省科技研发计划联合基金应用攻关项目（232103810043）；河南省高校科技创新人才支持计划资助（24HASTIT043）；河南省高等学校青年骨干教师培养计划（2021GGJS048）；河南省青年托举人才项目（2023HYTP050）

详细信息

作者简介:
李军星（1990-），男，副教授，博士，研究领域为可靠性设计与评估、剩余寿命预测。E-mail：lijunxing@haust.edu.cn

中图分类号: V229.2；TH133.33
计量
- 文章访问数: 432
- HTML浏览量: 222
- PDF量: 37
- 被引次数: 0
出版历程
- 收稿日期: 2023-11-07
- 网络出版日期: 2024-05-17

Life prediction method of rolling bearings in noisy environments based on IKF-ARIMA-NARNN model

LI Junxing^{1, 2, 3
,},
FAN Jiahui^1
,,
WANG Zhihua⁴,
FU Huimin⁴

1.
School of Mechatronical Engineering，Henan University of Science and Technology，Luoyang Henan 471000，China
2.
Advanced Manufacturing of Machinery and Equipment Collaborative Innovation Center，Henan University of Science and Technology，Luoyang Henan 471000，China
3.
High-end Bearing Collaborative Innovation Center of Henan Province，Henan University of Science and Technology，Luoyang Henan 471000，China
4.
Research Center of Small Sample Technology，Beihang University，Beijing 100191，China

摘要

摘要:
针对滚动轴承振动信号性能特征参数中不可避免地受环境噪声影响，同时又不具备马尔科夫特性的情况，提出基于IKF-ARIMA-NARNN的滚动轴承噪声环境下寿命预测方法。首先，考虑到传统卡尔曼滤波（KF）去噪忽略了数据间相关性的问题，提出一种基于增量卡尔曼滤波（IKF）的滚动轴承性能特征数据去噪方法。其次，针对滚动轴承性能特征参数演化过程有时不满足马尔科夫特性，建立基于具有非线性误差项的自回归差分移动平均（ARIMA）模型的滚动轴承性能特征参数演化过程分析模型；同时，利用动态非线性自回归神经网络（NARNN）估计退化模型非线性随机误差项，从而实现滚动轴承寿命预测。最后，通过滚动轴承工程实例分析，验证了该文方法的有效性和适用性，与传统方法相比，该方法的预测精度至少提高26.75%和51.25%。
- 滚动轴承 /
- 噪声环境 /
- 寿命预测 /
- 非线性自回归神经网络（NARNN） /
- 自回归差分移动平均（ARIMA） /
- 增量卡尔曼滤波（IKF）
Abstract:
In view of the fact that the performance characteristics of rolling bearings vibration signals are inevitably affected by environmental noise and lack of Markov characteristics, a life prediction method of rolling bearings in noisy environment based on IKF-ARIMA-NARNN was proposed. Firstly, considering that the Kalman filter （KF） denoising ignored the correlation of data, a denoising method based on incremental Kalman filter （IKF） was proposed. Secondly, due to the evolution process of bearing performance characteristic parameters that fails to meet the Markov characteristics sometimes, an analysis model of the evolution process of bearing performance characteristic parameters based on the autoregressive integrated moving average （ARIMA） with nonlinear error term was established. At the same time, the dynamic nonlinear autoregressive neural network （NARNN） was used to estimate the nonlinear random error term of the degradation model, so as to realize the life prediction of rolling bearings. Finally, the effectiveness and applicability of the proposed method was verified by the case analysis of rolling bearing engineering, and the prediction accuracy was at least 26.75% and 51.25% higher than the traditional method.
- rolling bearing /
- noisy environment /
- life prediction /
- nonlinear autoregressive neural network （NARNN） /
- autoregressive integrated moving average （ARIMA） /
- incremental Kalman filter （IKF）

HTML

图 1 NARNN网络结构图

Figure 1. NARNN network structure diagram

下载: 全尺寸图片幻灯片

图 2 NARNN开环模式

Figure 2. NARNN open-loop model

下载: 全尺寸图片幻灯片

图 3 NARNN闭环模式

Figure 3. NARNN closed-loop model

下载: 全尺寸图片幻灯片

图 4 寿命预测流程图

Figure 4. Life prediction flowchart

下载: 全尺寸图片幻灯片

图 5 滚动轴承寿命试验

Figure 5. Life test of rolling bearing

下载: 全尺寸图片幻灯片

图 6 3种去噪方法处理结果

Figure 6. Comparison of the processing results of the three denoising methods

下载: 全尺寸图片幻灯片

图 7 1号轴承性能特征参数预测对比

Figure 7. Comparison of performance characteristics predictions of bearing 1

下载: 全尺寸图片幻灯片

图 8 2号轴承性能特征参数预测对比

Figure 8. Comparison of performance characteristics predictions of bearing 2

下载: 全尺寸图片幻灯片

表 1 3种去噪方法的SNR和MSE对比

Table 1. Comparison of SNR and MSE of the three denoising methods

方法	信噪比	均方误差
卡尔曼滤波	16.4448	0.6726
小波阈值	26.9660	0.0597
增量卡尔曼滤波	29.8079	0.0390

下载: 导出CSV

表 2 1号轴承预测MAE和MRE对比

Table 2. Comparison of MAE and MRE of bearing 1

模型	MAE	MRA
RBFNN	3.8464	0.5857
NARNN	0.7239	0.0999
ARIMA-NARNN	0.5662	0.0801
IKF-ARIMA-NARNN	0.3477	0.0582

下载: 导出CSV

表 3 2号轴承预测MAE和MRE对比

Table 3. Comparison of MAE and MRE of bearing 2

模型	MAE	MRE
RBFNN	2.3131	0.3363
NARNN	0.7078	0.0971
ARIMA-NARNN	0.5317	0.0848
IKF-ARIMA-NARNN	0.3094	0.0518

下载: 导出CSV

表 4 寿命预测对比

Table 4. Comparison of life predictions

模型	1号		2号
模型	$\hat l_{{k}} $/min	绝对误差	$\hat l_{{k}} $/min	绝对误差
Wiener	12110.44	5.44	5329.79	9.21
NARNN	12090.44	14.56	5354.76	15.76
ARIMA-NARNN	12109.01	4.01	5345.13	6.13
IKF-ARIMA-NARNN	12106.53	1.53	5343.49	4.49

下载: 导出CSV

参考文献(22)

[1]	中国轴承工业协会. 高端轴承技术路线图[M]. 北京: 中国科学技术出版社,2018. CHINA Bearing Industry Association. High-end bearings technology roadmaps[M]. Beijing: China Science and Technology Press,2018. (in Chinese CHINA Bearing Industry Association. High-end bearings technology roadmaps[M]. Beijing: China Science and Technology Press, 2018. (in Chinese)
[2]	PRAKASH G,YUAN X,HAZRA B,et al. Toward a big data-based approach: a review on degradation models for prognosis of critical infrastructure[J]. Journal of Nondestructive Evaluation Diagnostics and Prognostics of Engineering Systems,2021,4(2): 021005. doi: 10.1115/1.4048787
[3]	WEN Y,RAHMAN M F,XU H,et al. Recent advances and trends of predictive maintenance from data-driven machine prognostics perspective[J]. Measurement,2022,187: 112076.
[4]	ZHANG P,GAO Z,CAO L,et al. Marine systems and equipment prognostics and health management: a systematic review from health condition monitoring to maintenance strategy[J]. Machines,2022,10(2): 72. doi: 10.3390/machines10020072
[5]	MADAR E,KLEIN R,BORTMAN J. Contribution of dynamic modeling to prognostics of rotating machinery[J]. Mechanical Systems and Signal Processing,2019,123: 496-512. doi: 10.1016/j.ymssp.2019.01.003
[6]	AIVALIOTIS P,GEORGOULIAS K,ARKOULI Z,et al. Methodology for enabling digital twin using advanced physics-based modelling in predictive maintenance[J]. Procedia CIRP,2019,81: 417-422. doi: 10.1016/j.procir.2019.03.072
[7]	YU W A,HARRIS T A. A new stress-based datigue life model for ball Bearings[J]. Tribology Transactions,2021,44(1): 11-18.
[8]	LEI Y,LI N,GONTARZ S,et al. A Model based method for remaining useful life prediction of machinery[J]. IEEE Transactions on Reliability,2016,65(3): 1314-1326. doi: 10.1109/TR.2016.2570568
[9]	RAMEZANI S,MOINI A,RIAHI M. Prognostics and health management in machinery: a review of methodologies for RUL prediction and roadmap[J]. International Journal of Industrial Engineering and Manage ment Science,2019,6(1): 38-61.
[10]	SONG Y,XU S,LU X. A sliding sequence importance resample filtering method for rolling bearings remaining useful life prediction based on two Wiener-process models[J]. Measurement Science and Technology,2024,35(1): 015019. doi: 10.1088/1361-6501/acffe3
[11]	GUAN Q,WEI X,ZHANG H,et al. Remaining useful life prediction for degradation processes based on the Wiener process considering parameter dependence[J]. Quality and Reliability Engineering International,2024,40(3): 1221-1245. doi: 10.1002/qre.3461
[12]	LI Y,HUANG X,DING P,et al. Wiener-based remaining useful life prediction of rolling bearings using improved Kalman filtering and adaptive modification[J]. 2021,182: 109706
[13]	GE R,ZHAI Q,WANG H,et al. Wiener degradation models with scale-mixture normal distributed measurement errors for RUL prediction[J]. Mechanical Systems and Signal Processing,2022,173: 109029. doi: 10.1016/j.ymssp.2022.109029
[14]	CHENG Y,ZHU H,HU K,et al. Reliability prediction of machinery with multiple degradation characteristics using double-Wiener process and Monte Carlo algorithm[J]. Mechanical Systems and Signal Processing,2019,134: 106333. doi: 10.1016/j.ymssp.2019.106333
[15]	LI N,NAGI G,LEI Y,et al. Remaining useful life prediction of machinery under time-varying operating conditions based on a two-factor state-space model[J]. Reliability Engineering & System Safety,2019,186: 88-100.
[16]	WANG H,MA X,ZHAO Y. An improved Wiener process model with adaptive drift and diffusion for online remaining useful life prediction[J]. Mechanical Systems and Signal Processing,2019,127: 370-387. doi: 10.1016/j.ymssp.2019.03.019
[17]	PENG Y,WANG Y,ZI Y. Switching state-space degradation model with recursive filter/smoother for prognostics of remaining useful life[J]. IEEE Transactions on Industrial Informatics,2019,15(2): 822-832. doi: 10.1109/TII.2018.2810284
[18]	WANG Y,PENG Y,ZI Y,et al. A two-stage data-driven-based prognostic approach for bearing degradation problem[J]. IEEE Transactions on Industrial Informatics,2016,12(3): 924-932. doi: 10.1109/TII.2016.2535368
[19]	FENG L,DU L,GUO J,et al. A bias drift suppression method based on ICELMD and ARMA-KF for MEMS gyros[J]. Micromachines,2022,14(1): 109. doi: 10.3390/mi14010109
[20]	LI J,CHEN W,HAN K,et al. Fault diagnosis of rolling bearing based on GA-VMD and improved WOA-LSSVM[J]. IEEE Access,2020,8: 166753-166767. doi: 10.1109/ACCESS.2020.3023306
[21]	HAMDAOUI H,NGIEJUNGBWEN L A,GU J,et al. Improved signal processing for bearing fault diagnosis in noisy environments using signal denoising,time-frequency transform,and deep learning[J]. Journal of the Brazilian Society of Mechanical Sciences and Engineering,2023,45(11): 576. doi: 10.1007/s40430-023-04471-9
[22]	傅惠民,娄泰山,吴云章. 欠观测条件下的扩展增量Kalman滤波方法[J]. 航空动力学报,2012,27(4): 777-781. FU Huimin,LOU Taishan,WU Yunzhang. Extended incremental Kalman filter method under poor observation condition[J]. Journal of Aerospace Power,2012,27(4): 777-781. (in Chinese FU Huimin, LOU Taishan, WU Yunzhang. Extended incremental Kalman filter method under poor observation condition[J]. Journal of Aerospace Power, 2012, 27(4): 777-781. (in Chinese)