Fault diagnosis of high speed ball bearing based on elastic kernel convex hull tensors
-
摘要: 为了提高机械传动系统故障诊断精度，设计了一种基于弹性核凸包张量机（flexible convex hull tensor machine, FCHTM）的轴承故障诊断方法。采用当前具备成熟技术的连续小波转换方法建立时频分布，再根据时频图灰度共生矩阵数据建立特征集，由此计算得到精确的时频能量谱纹理特征。利用训练弹性核凸包张量机模型来识别测试集样本，实现故障的快速诊断。研究结果表明：采用本文方法可以获得比初始弹性凸包分类方式更高准确率，时频能量谱纹理特征相对传统时频特征达到了更高准确率。弹性核凸包张量机具备优异泛化性能，达到了优分类精度。相比较支持向量机（support vector machine, SVM）和弹性核凸包（flexible convex hull, FCH），采用弹性核凸包张量机方法达到了更高准确率，弹性核凸包张量机表现出了更优抗噪性与鲁棒性。Abstract: In order to improve the fault diagnosis accuracy of mechanical transmission system, a fault diagnosis method of high-speed ball bearing based on elastic kernel convex hull tensor was designed. The time-frequency distribution is established by using the continuous wavelet transform method with mature technology, and then the feature set is established according to the data of the gray co-occurrence matrix of the time-frequency graph, from which the accurate texture features of the time-frequency graph are calculated. The model of training elastic kernel convex hull tensor is used to identify test set samples and realize fast fault diagnosis. The research results show that the proposed method can obtain higher accuracy than the initial elastic convex hull classification method, and the texture features of time-frequency maps achieve higher accuracy than the traditional time-frequency features. The elastic kernel convex hull tensor has excellent generalization performance and achieves the optimal classification accuracy. The training set and test set were divided by the ratio of 8∶2, and the elastic kernel convex hull tensor was used to achieve the highest accuracy. The elastic kernel convex hull tensor showed better anti-noise and robustness. The research can be applied to other mechanical transmission systems and has a good value for practical promotion.
-
Key words:
- probability output /
- elastic convex hull /
- high speed ball bearing /
- fault diagnosis
-
表 1 轴承数据样本
数据集 故障尺寸/μm 故障位置 样本数 标签 A 0 120 1 216 滚动体 120 2 384 滚动体 120 3 596 滚动体 120 4 216 外圈 120 5 384 外圈 120 6 596 外圈 120 7 216 内圈 120 8 384 内圈 120 9 596 内圈 120 10 B 0 120 1 216/384/596 滚动体 480 2 216/384/596 外圈 480 3 216/384/596 内圈 480 4 
下载: 导出CSV
表 2 识别准确率
模型 准确率/(%) 数据集A 数据集B SVM 89.96±0.12 91.42±0.13 FCH 92.82±0.19 93.02±0.14 FCHTM 98.46±0.12 97.85±0.06 TEST+SVM 96.25±0.11 95.94±0.11 TEST+FCH 99.32±0.07 99.21±0.05 TEST+FCHTM 99.78±0.05 99.81±0.03
下载: 导出CSV
表 4 识别准确率及运行时间
模型 识别精度/(%) 训练时间/s 测试时间/s SVM 92.12±0.26 0.022 3 0.000 6 FCH 95.62±0.15 0.039 6 0.001 7 FCHTM 98.78±0.42 0.046 2 0.002 2 TEST+SVM 95.16±0.19 0.010 6 0.000 4 TEST+FCH 99.02±0.21 0.038 5 0.001 3 TEST+FCHTM 99.92±0.16 0.044 1 0.001 6
下载: 导出CSV
表 5 鲁棒性测试结果
数据集 SVM FCH FCHTM A 识别精度/(%) 94.51±0.32 96.05±0.26 98.85±0.35 识别精度
下降率/(%)4.16 3.75 2.62 B 识别精度/(%) 93.46±0.46 97.15±0.16 99.12±0.18 识别精度下
降率/(%)6.41 6.22 4.15
下载: 导出CSV
-
[1] 张昭晗, 齐俊平, 李峰, 等. 基于FCMMWPE-BSASVM组合算法的调心球轴承故障诊断研究[J]. 制造技术与机床, 2022(11): 15-19. doi: 10.19287/j.mtmt.1005-2402.2022.11.002 [2] He G, Ding K, Lin H. Gearbox coupling modulation separation method based on match pursuit and correlation filtering[J]. Mechanical Systems and Signal Processing, 2016, 66: 597-611. [3] 张伟, 李军霞, 陈维望. 基于蝙蝠算法优化VMD参数的滚动轴承复合故障分离方法[J]. 振动与冲击, 2022, 41(20): 133-141. doi: 10.13465/j.cnki.jvs.2022.20.017 [4] 肖俊青, 岳敏楠, 李春, 等. 基于固有时间尺度分解与卷积神经网络的轴承故障诊断研究[J]. 机械强度, 2022, 44(5): 1017-1023. [5] Yang X Q, Ding K, He G L, et al. Double-dictionary signal decomposition method based on split augmented Lagrangian shrinkage algorithm and its application in gearbox hybrid faults diagnosis[J]. Journal of Sound and Vibration, 2018, 432: 484-501. doi: 10.1016/j.jsv.2018.06.064 [6] 包从望, 江伟, 郭灏, 等. 基于低秩稀疏分解的采煤机轴承复合故障分离[J]. 中国工程机械学报, 2022, 20(4): 364-370. [7] 邱海枫, 苏宁, 田松林. 改进支持向量机在电力变压器故障诊断中的应用研究[J]. 电测与仪表, 2022, 59(11): 48-53. doi: 10.19753/j.issn1001-1390.2022.11.007 [8] 徐乐, 朱玉斌, 郎超男. 基于LMD能量熵和支持向量机的齿轮箱故障诊断[J]. 制造技术与机床, 2022(6): 44-49. [9] 周俊宏, 徐睿, 张惠荣, 等. 自适应蝴蝶算法优化DBN的无人直升机电机轴承故障诊断[J]. 机械设计与研究, 2022, 38(2): 124-127. [10] 王勉, 刘勇. 基于时移多尺度散布熵和SVM的滚动轴承故障诊断方法[J]. 机械设计与研究, 2021, 37(5): 83-87. [11] 马怀祥, 冯旭威, 李东升, 等. 基于CNN和XGBoost的滚动轴承故障诊断方法[J]. 中国工程机械学报, 2021, 19(3): 254-259. [12] Djebala A, Ouelaa N, Benchaabane C, et al. Application of the wavelet multi-resolution analysis and Hilbert transform for the prediction of gear tooth defects[J]. Meccanica, 2012, 47(7): 1601-1612. doi: 10.1007/s11012-012-9538-1 [13] Fan X F, Zuo M J. Gearbox fault detection using Hilbert and wavelet packet transform[J].Mechanical Systems and Signal Processing, 2006, 20(4): 966-982. [14] 何知义, 邵海东, 程军圣, 等. 基于弹性核凸包张量机的机械设备热成像故障诊断方法[J]. 中国机械工程, 2021, 32(12): 1456-1461, 1478. [15] 杨路航, 李宝庆, 王平, 等. 基于概率输出弹性凸包的滚动轴承故障诊断方法[J]. 中国机械工程, 2021, 32(1): 40-46. [16] 李红光, 郭英, 眭萍, 等. 基于时频能量谱纹理特征的跳频调制方式识别[J]. 通信学报, 2019, 40(10): 20-29. doi: 10.11959/j.issn.1000-436x.2019191 -
下载:
点击查看大图
图(3) / 表(5)
计量
- 文章访问数: 715
- HTML全文浏览量: 32
- PDF下载量: 32
- 被引次数: 0
