通信学报 ›› 2024, Vol. 45 ›› Issue (4): 185-200.doi: 10.11959/j.issn.1000-436x.2024081
• 学术通信 • 上一篇
黑新宏1,2, 高苗1, 张宽1, 费蓉1,2, 邱原1,2, 姬文江1,2
收稿日期:
2023-12-07
修回日期:
2024-03-12
出版日期:
2024-04-30
发布日期:
2024-05-27
作者简介:
基金资助:
Xinhong HEI1,2, Miao GAO1, Kuan ZHANG1, Rong FEI1,2, Yuan QIU1,2, Wenjiang JI1,2
Received:
2023-12-07
Revised:
2024-03-12
Online:
2024-04-30
Published:
2024-05-27
Supported by:
摘要:
为了提高故障诊断模型在数据不平衡场景下的诊断性能和模型泛化能力,提出了一种基于Nadam-TimeGAN和XGBoost的时序信号故障诊断方法。首先对比基于LSTM和GRU的TimeGAN模型,选取性能更优的GRU网络作为TimeGAN模型的组成单元,然后采用Nadam优化算法对TimeGAN模型的各组件进行优化,即构建Nadam-TimeGAN模型用以数据扩充,最后构建一个平衡的数据集输入XGBoost集成学习模型进行分类训练。实验选取转辙机动作电流数据集进行验证性实验,选取MFPT轴承数据集和CWRU轴承数据集进行泛化性实验,并与8种方法进行对比,结果表明,所提方法在准确率、召回率以及F1-score这3种评价指标上均高于其他方法,从而验证了所提方法在不平衡数据故障诊断方面的有效性和泛化性。
中图分类号:
黑新宏, 高苗, 张宽, 费蓉, 邱原, 姬文江. 基于Nadam-TimeGAN和XGBoost的时序信号故障诊断方法[J]. 通信学报, 2024, 45(4): 185-200.
Xinhong HEI, Miao GAO, Kuan ZHANG, Rong FEI, Yuan QIU, Wenjiang JI. Fault diagnosis method of timing signal based on Nadam-TimeGAN and XGBoost[J]. Journal on Communications, 2024, 45(4): 185-200.
表10
6类实验的所提方法与其他方法对比结果"
实验 | 方法 | 准确率 | 召回率 | F1-score | |
---|---|---|---|---|---|
a | Nadam- TimeGAN | XGBoost | 99.21% | 99.21% | 0.992 2 |
LightGBM | 98.81% | 98.81% | 0.988 2 | ||
GBDT | 98.74% | 98.74% | 0.987 5 | ||
b | TimeGAN- GRU | XGBoost | 98.67% | 98.67% | 0.986 3 |
LightGBM | 97.91% | 97.90% | 0.979 2 | ||
GBDT | 98.04% | 98.03% | 0.980 5 | ||
c | TimeGAN- LSTM | XGBoost | 93.21% | 93.21% | 0.932 2 |
LightGBM | 92.29% | 92.27% | 0.922 8 | ||
GBDT | 92.67% | 92.67% | 0.926 3 | ||
d | 原始不平衡数据集 | XGBoost | 94.91% | 94.55% | 0.946 4 |
LightGBM | 94.66% | 93.63% | 0.944 2 | ||
GBDT | 94.25% | 93.91% | 0.943 2 | ||
e | 原始平衡数据集 | XGBoost | 96.76% | 96.19% | 0.961 5 |
LightGBM | 96.09% | 96.30% | 0.962 8 | ||
GBDT | 95.43% | 95.66% | 0.956 0 | ||
f | Nadam- TimeGAN | XGBoost | 95.43% | 95.78% | 0.957 6 |
LightGBM | 95.84% | 95.08% | 0.951 7 | ||
GBDT | 95.38% | 95.30% | 0.953 3 |
表13
所提方法与其他方法在MFPT数据集对比结果"
实验 | 方法 | 准确率 | 召回率 | F1-score | |
---|---|---|---|---|---|
a | Nadam- TimeGAN | XGBoost | 97.99% | 97.99% | 0.979 8 |
LightGBM | 97.91% | 97.89% | 0.978 9 | ||
GBDT | 97.38% | 97.36% | 0.973 6 | ||
b | TimeGAN- GRU | XGBoost | 97.81% | 97.81% | 0.978 6 |
LightGBM | 97.24% | 97.30% | 0.973 1 | ||
GBDT | 96.22% | 96.19% | 0.962 0 | ||
c | TimeGAN- LSTM | XGBoost | 96.95% | 96.95% | 0.970 0 |
LightGBM | 96.80% | 96.73% | 0.967 4 | ||
GBDT | 96.12% | 96.09% | 0.960 4 | ||
d | 原始不平衡数据集 | XGBoost | 93.54% | 93.53% | 0.935 2 |
LightGBM | 93.15% | 93.22% | 0.932 1 | ||
GBDT | 92.95% | 92.95% | 0.929 3 | ||
e | 原始平衡数据集 | XGBoost | 97.19% | 97.07% | 0.970 8 |
LightGBM | 97.06% | 96.94% | 0.969 5 | ||
GBDT | 96.35% | 96.11% | 0.962 6 | ||
f | Nadam- TimeGAN | XGBoost | 96.92% | 95.37% | 0.953 0 |
LightGBM | 96.62% | 96.46% | 0.964 9 | ||
GBDT | 96.22% | 96.07% | 0.960 8 |
表16
不同分类方法在CWRU数据集上对比结果"
实验 | 方法 | 准确率 | 召回率 | F1-score | |
---|---|---|---|---|---|
a | Nadam- TimeGAN | XGBoost | 96.34% | 96.50% | 0.965 6 |
LightGBM | 96.04% | 96.19% | 0.962 9 | ||
GBDT | 95.83% | 95.87% | 0.958 5 | ||
b | TimeGAN | XGBoost | 95.99% | 96.13% | 0.960 7 |
LightGBM | 95.47% | 95.61% | 0.955 7 | ||
GBDT | 94.66% | 94.84% | 0.947 4 | ||
c | 原始不平衡数据集 | XGBoost | 92.56% | 92.60% | 0.925 8 |
LightGBM | 91.50% | 91.53% | 0.915 0 | ||
GBDT | 91.08% | 91.08% | 0.908 6 | ||
d | 原始平衡数据集 | XGBoost | 95.08% | 95.13% | 0.951 0 |
LightGBM | 94.49% | 94.41% | 0.942 8 | ||
GBDT | 94.49% | 94.19% | 0.941 3 | ||
e | Nadam- TimeGAN | XGBoost | 94.74% | 94.54% | 0.941 9 |
LightGBM | 94.33% | 94.13% | 0.941 0 | ||
GBDT | 94.22% | 93.21% | 0.933 7 | ||
f | SMOTE | XGBoost | 92.71% | 92.72% | 0.927 4 |
LightGBM | 91.68% | 91.74% | 0.917 4 | ||
GBDT | 92.07% | 92.04% | 0.920 1 | ||
g | DCGAN-CNN | 93.98% | 93.94% | 0.939 1 | |
h | MLP | 89.55% | 89.63% | 0.895 5 | |
i | FCN | 87.69% | 86.07% | 0.837 6 |
1 | 来杰, 王晓丹, 向前, 等. 自编码器及其应用综述[J]. 通信学报, 2021, 42(9): 218-230. |
LAI J, WANG X D, XIANG Q, et al. Review on autoencoder and its application[J]. Journal on Communications, 2021, 42(9): 218-230. | |
2 | FEI R, WAN Y X, HU B, et al. A novel network core structure extraction algorithm utilized variational autoencoder for community detection[J]. Expert Systems with Applications, 2023, 222: 119775. |
3 | COULIBALY S, KAMSU-FOGUEM B, KAMISSOKO D, et al. Deep convolution neural network sharing for the multi-label images classification[J]. Machine Learning with Applications, 2022, 10: 100422. |
4 | GHOFRANI F, SUN H Y, HE Q. Analyzing risk of service failures in heavy haul rail lines: a hybrid approach for imbalanced data[J]. Risk Analysis, 2022, 42(8): 1852-1871. |
5 | GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets[J]. Advances in Neural Information Processing Systems, 2014, 27: 1-9. |
6 | RUAN D W, SONG X Z, GüHMANN C, et al. Collaborative optimization of CNN and GAN for bearing fault diagnosis under unbalanced datasets[J]. Lubricants, 2021, 9(10): 105. |
7 | MA J Q, JIANG X X, HAN B K, et al. Dynamic simulation model-driven fault diagnosis method for bearing under missing fault-type samples[J]. Applied Sciences, 2023, 13(5): 2857. |
8 | WANG R G, CHEN Z Y, ZHANG S H, et al. Dual-attention generative adversarial networks for fault diagnosis under the class-imbalanced conditions[J]. IEEE Sensors Journal, 2022, 22(2): 1474-1485. |
9 | WANG R G, ZHANG S H, CHEN Z Y, et al. Enhanced generative adversarial network for extremely imbalanced fault diagnosis of rotating machine[J]. Measurement, 2021, 180: 109467. |
10 | YOON J, JARRETT D, SCHAAR M. Time-series generative adversarial networks[J]. Advances in Neural Information Processing Systems, 2019, 32: 1-11. |
11 | 黄开远, 罗娜. 基于Attention机制改进TimeGAN的小样本时间序列预测方法[J]. 华东理工大学学报(自然科学版), 2023, 49(6): 890-899. |
HUANG K Y, LUO N. Improved TimeGAN based on Attention for time series prediction method with few shot[J]. Journal of East China University of Science and Technology, 2023, 49(6): 890-899. | |
12 | ZHONG S P, WANG X M, XU B, et al. Hybrid network model based on data enhancement for short-term power prediction of new PV plants[J]. Journal of Modern Power Systems and Clean Energy, 2024, 12(1): 77-88. |
13 | SHI Y L, LI J C, LI H G, et al. An imbalanced data augmentation and assessment method for industrial process fault classification with application in air compressors[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 3521510. |
14 | SHI H B, XU Y P, DING B D, et al. Long-term solar power time-series data generation method based on generative adversarial networks and sunrise–sunset time correction[J]. Sustainability, 2023, 15(20): 14920. |
15 | LU J, LIU Z L, ZHANG W J, et al. Pressure prediction study of coal mining working face based on nadam-LSTM[J]. IEEE Access, 2023, 11: 83867-83880. |
16 | SUN X L, XIE M K, ZHOU F, et al. Hierarchical evolutionary construction of neural network models for an Atkinson cycle engine with double injection strategy based on the PSO-Nadam algorithm[J]. Fuel, 2023, 333: 126531. |
17 | QIAO M Y, TANG X X, SHI J K, et al. Bearing fault diagnosis based on natural adaptive moment estimation algorithm and improved octave convolution[J]. IEEE Access, 2020, 8: 196790-196803. |
18 | 李占山, 刘兆赓. 基于XGBoost的特征选择算法[J]. 通信学报, 2019, 40(10): 101-108. |
LI Z S, LIU Z G. Feature selection algorithm based on XGBoost[J]. Journal on Communications, 2019, 40(10): 101-108. | |
19 | HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780. |
20 | FANG W, ZHANG F H, SHENG V S, et al. A method for improving CNN-based image recognition using DCGAN[J]. Computers, Materials & Continua, 2018, 57(1): 167-178. |
21 | 司念文, 张文林, 屈丹, 等. 卷积神经网络表征可视化研究综述[J]. 自动化学报, 2022, 48(8): 1890-1920. |
SI N W, ZHANG W L, QU D, et al. Representation visualization of convolutional neural networks: a survey[J]. Acta Automatica Sinica, 2022, 48(8): 1890-1920. | |
22 | 周翰逊, 陈晨, 冯润泽, 等. 基于值导数GRU的移动恶意软件流量检测方法[J]. 通信学报, 2020, 41(1): 102-113. |
ZHOU H X, CHEN C, FENG R Z, et al. Mobile malware traffic detection approach based on value-derivative GRU[J]. Journal on Communications, 2020, 41(1): 102-113. | |
23 | DANG B L, NGUYEN-XUAN H, WAHAB M A. An effective approach for VARANS-VOF modelling interactions of wave and perforated breakwater using gradient boosting decision tree algorithm[J]. Ocean Engineering, 2023, 268: 113398. |
24 | 朱晓荣, 张佩佩. 基于GAN的异构无线网络故障检测与诊断算法[J]. 通信学报, 2020, 41(8): 110-119. |
ZHU X R, ZHANG P P. Fault detection and diagnosis method for heterogeneous wireless network based on GAN[J]. Journal on Communications, 2020, 41(8): 110-119. |
[1] | 艾浩军, 曾维珂, 陶荆杰, 徐锦盈, 常含笑. 基于扩散模型的室内定位射频指纹数据增强方法[J]. 通信学报, 2023, 44(11): 201-212. |
[2] | 汤凌韬, 王迪, 刘盛云. 面向非独立同分布数据的联邦学习数据增强方案[J]. 通信学报, 2023, 44(1): 164-176. |
[3] | 冯晓伟, 许剑锋, 何川. 动态广义主成分分析及其在故障子空间建模中的应用[J]. 通信学报, 2022, 43(5): 92-101. |
[4] | 言洪萍, 周强, 王世豪, 姚旺, 何刘坤, 王良民. 基于SDN的实际网络流中Tor网页复合特征提取方法[J]. 通信学报, 2022, 43(3): 76-87. |
[5] | 朱晓荣,张佩佩. 基于GAN的异构无线网络故障检测与诊断算法[J]. 通信学报, 2020, 41(8): 110-119. |
[6] | 陈哲,胡玉其,田世庆,陆慧敏,徐立中. 基于非平稳信号组合分析的故障诊断方法[J]. 通信学报, 2020, 41(5): 187-195. |
[7] | 汪晓丁,许力,周书明,林晖. 基于Voronoi图和定位集的故障诊断和修复策略[J]. 通信学报, 2015, 36(Z1): 103-109. |
[8] | 黄静,成璐,邱雪松,龙会湖. Internet服务故障的影响分析模型[J]. 通信学报, 2008, 29(12): 73-80. |
[9] | 陈鸣,周骏,常强林,高屹. 用移动探针构造端到端故障诊断系统[J]. 通信学报, 2006, 27(5): 100-106. |
[10] | 朱有产,窦炳琳,商李彪,余晓晔. 网络故障诊断框架的研究及应用[J]. 通信学报, 2005, 26(1A): 68-71. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||
|