Broad transfer learning algorithm based on angular domain resampling
-
摘要: 针对在变工况中采集到的信号通常为非平稳信号，并且采集到的振动数据分布不一致等情况，提出了基于角域重采样的宽度迁移学习（ADR-BTL）算法，用于轴承故障诊断。首先将非平稳的时域振动信号转换为角域平稳信号，将平稳化处理后的信号进行特征提取，构建多域特征数据集，然后将不同工况下的源域数据和目标域数据通过平衡分布自适应（BDA）方法进行领域适配来减小域间的分布差异，最后构建宽度迁移学习模型。实验首先验证了角域重采样方法可以将振动信号进行平稳化处理，然后通过仿真样本分析得出BDA方法能够解决数据分布不一致问题，最后通过实验结果得出，提出的ADR-BTL识别率达到了98.9%，识别效果是最好的，证明了所提的方法在轴承故障诊断方面是有效的。Abstract: An broad transfer learning algorithm based on angular domain resampling(ADR-BTL) is proposed for bearing fault diagnosis in response to the fact that the signals collected in variable operating conditions are usually non-stationary signals and the distribution of the collected vibration data is inconsistent. Firstly, the non-stationary time domain vibration signal is converted into an angular domain stationary signal, and the smoothed signal is feature extracted to construct a multi domain feature dataset. Then, the source domain data and target domain data under different operating conditions are domain adapted using the balanced distribution adaptation (BDA) method to reduce the distribution differences between domains. Finally, a broad transfer learning model is constructed. The experiment first verified that the angular domain resampling method can smooth the vibration signal, and then through simulation sample analysis, it was found that the BDA method can solve the problem of inconsistent data distribution. Finally, the experimental results showed that the proposed ADR-BTL recognition rate reached 98.9%, and the recognition effect was the best, proving that the proposed method is effective in bearing fault diagnosis.
-
算法：ADR-BTL 输入：源域样本{(xs, ys)|xs∈Rn*D, ys∈${{{\bf{R}}^{{\boldsymbol{n}} * {C_0}}}} $}，目标域样本{xt∈Rm*D}， 宽度学习的正则化参数θ，最大迭代次数Tmax 输出：目标域数据的识别结果步骤1：初始化迭代次数T=0，BDA方法的平衡因子μ=0步骤2：通过式（22）与式（23）初始化源域和目标域构成的 MMD矩阵M0，Mc步骤3：基于源域数据xs训练1个基分类器，预测xt对应的伪标签$ {\overline {{y_t}}} $步骤4：while ${ T \leqslant {T_{\max }} }$ 利用适配后的数据{ATxs，ys}训练得到分类器f 更新伪标签：${ \overline {{y_t}} = f({{\boldsymbol{A}}^{\rm{T}}}{{{\boldsymbol{x}}_{t}}}) }$ 通过式（23）更新矩阵Mc 利用式（13）更新平衡因子μ T=T+1 End步骤5：求解式（24），获得最优的变换矩阵A步骤6：通过变换矩阵A得到映射后的源域样本ATxs和目标域样本 ATxt 步骤7：构建特征节点Zi，然后通过非线性变换为增强节点Hj，将 Zi和Hj组合输出步骤8：构建宽度迁移学习模型步骤9：输出目标域数据的识别结果 
下载: 导出CSV
-
[1] 宫涛, 杨建华, 单振, 等. 强噪声背景与变转速工况条件下滚动轴承故障诊断研究[J]. 工矿自动化,2021,47(7):63-71. doi: 10.13272/j.issn.1671-251x.17757 [2] 刘鲲鹏, 白云川, 吕麒鹏, 等. 基于角域功率谱估计的滚动轴承故障诊断[J]. 内燃机与配件,2018(19):88-89. doi: 10.19475/j.cnki.issn1674-957x.2018.19.041 [3] 吴剑, 代冀阳, 孙秀霞. 非平稳振动信号的角域重采样小波解调分析方法[J]. 振动与冲击,2009,28(5):137-141,159,210. doi: 10.3969/j.issn.1000-3835.2009.05.031 [4] Zhao R, Yan R, Chen Z, et al. Deep learning and its applications to machine health monitoring[J]. Mechanical Systems and Signal Processing,2019,115:213-237. doi: 10.1016/j.ymssp.2018.05.050 [5] Khan S, Yairi T. A review on the application of deep learning in system health management[J]. Mechanical Systems and Signal Processing,2018,107:241-265. doi: 10.1016/j.ymssp.2017.11.024 [6] Jia M, Xu Y, Hong M, et al. Multitask convolutional neural network for rolling element bearing fault identification[J]. Shock and Vibration,2020(2):1-10. [7] Li T F, Zhao Z B, Sun C, et al. WaveletKernelNet: An interpretable deep neural network for industrial intelligent diagnosis[J]. IEEE Transactions on Systems, Man, and Cybernetics:Systems,2021,52(4):2302-2312. [8] Chen Y, Peng G, Xie C, et al. ACDIN: Bridging the gap between artificial and real bearing damages for bearing fault diagnosis[J]. Neurocomputing,2018,294:61-71. doi: 10.1016/j.neucom.2018.03.014 [9] Wang L H, Zhao X P, Wu J X, et al. Motor fault diagnosis based on short-time Fourier transform and convolutional neural network[J]. Chinese Journal of Mechanical Engineering,2017,30(6):1357-1368. doi: 10.1007/s10033-017-0190-5 [10] Zhang Y, Xing K, Bai R, et al. An enhanced convolutional neural network for bearing fault diagnosis based on time–frequency image[J]. Measurement,2020,157:107667. doi: 10.1016/j.measurement.2020.107667 [11] 熊隽, 陈林. 注意力深度迁移联合适配的不同工况下旋转机械故障诊断方法[J]. 制造技术与机床,2022(10):29-37. [12] 段礼祥, 谢骏遥, 王凯, 等. 基于不同工况下辅助数据集的齿轮箱故障诊断[J]. 振动与冲击,2017,36(10):104-108,116. doi: 10.13465/j.cnki.jvs.2017.10.017 [13] 刘应东, 刘韬, 李华, 等. 变工况轴承的联合分布适应迁移故障诊断[J]. 电子测量与仪器学报,2021,35(5):69-75. doi: 10.13382/j.jemi.B2003511 [14] 王超. 基于宽度迁移学习网络的变负载下滚动轴承故障诊断方法研究[D]. 哈尔滨: 哈尔滨理工大学, 2021. [15] Wang J, Chen Y, Hao S, et al. Balanced distribution adaptation for transfer learning[C]// IEEE international conference on data mining (ICDM). IEEE, 2017: 1129-1134. [16] Chen C L P, Liu Z. Broad learning system: An effective and efficient incremental learning system without the need for deep architecture[J]. IEEE Transactions on Neural Networks and Learning Systems,2017,29(1):10-24. [17] 陈秋菊, 彭天昊, 张成龙, 等. 基于小波包分解和宽度学习的轴承故障检测[J]. 信息与电脑:理论版,2023,35(1):182-186. [18] 魏义敏, 刘辉, 杨乐红. 基于角域重采样与VMD的电梯曳引轮轴承故障诊断方法[J/OL]. 机电工程: 1-8[2023-03-30]. http://kns.cnki.net/kcms/ detail/33. 1088.th.20230316.1019.006.html -
下载:
点击查看大图
图(12) / 表(2)
计量
- 文章访问数: 47
- HTML全文浏览量: 4
- PDF下载量: 10
- 被引次数: 0
