基于收缩自编码器和局部保持投影的机械故障特征提取*

doi:10.16731/j.cnki.1671-3133.2025.02.011

摘要/Abstract

摘要： 局部保持投影算法的性能主要依赖于构造的最近邻图,而构造最近邻图时容易受到原始数据冗余信息的干扰,以及没有良好的依据选择合适的热核参数带来的影响,导致不能充分挖掘高维数据的局部结构信息,在低维嵌入过程中也易对噪声和异常值较为敏感,影响其在故障诊断应用中的特征提取能力。针对以上问题,提出基于收缩自编码器和流形排序的局部保持投影算法(Locality Preserving Projections algorithm based on Contractive Auto-Encoder and Manifold Ranking,CAE-MRLPP),并用于机械设备故障诊断。首先,将样本标签信息和斯皮尔曼相关系数结合,预调整样本间距;其次,引入流形排序思想,根据样本点与邻域点在彼此邻域集中的排序位置信息以及二者的互邻个数信息来构造权重;最后,将收缩自编码器与基于流形排序的局部保持投影相融合,通过梯度下降法迭代优化求解出最优的投影矩阵,进而得到故障数据的低维表示。分别在滚动轴承数据集和抽油机数据集上进行了多项验证,故障识别准确度均在98 %以上,表明该算法具有良好的特征提取能力,能够有效提高故障识别准确度,同时具有较好的鲁棒性和泛化能力。

关键词: 局部保持投影, 特征提取, 故障诊断, 收缩自编码器, 抽油机, 滚动轴承

Abstract: The performance of the locality preserving projections algorithm mainly depends on the construction of the nearest neighbor map. The construction of the nearest neighbor map was easily affected by the interference of redundant information in the original data and the impact of not having a good basis for selecting the appropriate heat kernel parameters.As a result,the local structural information of high-dimensional data cannot be fully explored,and it was also easy to be more sensitive to noise and outliers in the low-dimensional embedding process.And its feature extraction ability in fault diagnosis applications was affected. To address the above problems,a Locality Preserving Projections algorithm based on Contractive Auto-Encoder and Manifold Ranking (CAE-MRLPP) was proposed and used for mechanical equipment fault diagnosis. Firstly,it combined the label information and Spearman correlation coefficient to pre-adjust the sample spacing. Secondly,the idea of manifold ranking was introduced to construct the weights based on the information of the sorting position of the sample points and the neighboring points in each other′s neighborhood sets and the information of the number of mutual neighbors of the two. Lastly,the contractive auto-encoder was fused with the locality preserving projections based on the manifold ranking,and the optimal projection matrix was solved by iterative optimization of the gradient descent method.Then the low-dimensional representation of the fault data was obtained. A number of verifications were carried out on the rolling bearing dataset and the pumping unit dataset,and the fault identification accuracy was more than 98 %. This indicated that the algorithm has good feature extraction ability,can effectively improve the fault identification accuracy,and has good robustness and generalization ability.

Key words: locality preserving projections, feature extraction, fault diagnosis, contractive auto-encoder, oil pumping machine, rolling bearing

中图分类号:

TH165+.3

郝宇星, 刘庆强. 基于收缩自编码器和局部保持投影的机械故障特征提取^*[J]. 现代制造工程, 2025, 533(2): 84-93.

HAO Yuxing, LIU Qingqiang. Mechanical fault feature extraction based on contractive auto-encoder and locality preserving projections[J]. Modern Manufacturing Engineering, 2025, 533(2): 84-93.

参考文献

[1] KONGKLAD G,CHITAREE R,TAECHALERTPAISARN T,et al.Discriminant analysis pca-lda assisted surface-enhanced raman spectroscopy for direct identification of malaria-infected red blood cells[J].Methods and Protocols,2022,5(3):49-65.
[2] HE X,NIYOGI P. Locality preserving projections[J].Advances in Neural Information Processing Systems,2003,16(1):186-197.
[3] ROWEIS S T,SAUL L K. Nonlinear dimensionality reduction by Locally Linear Embedding[J].Science,2000,290:232.
[4] WANG J.Geometric structure of high-dimensional data and dimensionality reduction[M].[S.l.]:Springer and Chinese High-education Press,2012:221-234.
[5] BELKIN M,NIYOGI P.Laplacian eigenmaps and spectral techniques for embedding and clustering[J].Advances in Neural Information Processing Systems,2001,14(6):585-591.
[6] SU S Z,ZHU G,ZHU Y M,et al.Coupled locality discriminant analysis with globality preserving for dimensionality reduction[J].Applied Intelligence,2023,53(6):7118-7131.
[7] ZHANG N,TIAN Y,WANG X W,et al.Novel bootstrap-based discriminant NPE integrated with orthogonal LPP for fault diagnosis[J].IEEE Transactions on Instrumentation and Measurement,2023,72:1-9.
[8] WAN M H,YAO Y,ZHAN T M,et al.Supervised Low-Rank Embedded Regression (SLRER) for robust subspace learning[J].IEEE Transactions on Circuits and Systems for Video Technology,2021,32(4):1917-1927.
[9] ZHANG N,XU Y,ZHU Q X,et al.Improved locality preserving projections based on heat-kernel and cosine weights for fault classification in complex industrial processes[J].IEEE Transactions on Reliability,2022,72(1):204-213.
[10] GHOSH S,CHATTERJEE A,MUNSHI S.Visual cue-aided human supervised robot navigation guidance in photometrically challenging environments using adaptive spatial-feature kernel-guided bilateral LPP[J].Measurement Science and Technology,2023,34(10):105404.
[11] ZHANG N,XU Y,ZHU Q X,et al.Novel regularization double preserving integrated with neighborhood locality projections for fault diagnosis[J].IEEE Transactions on Industrial Informatics,2023,19(10):10478-10488.
[12] LU X H,LONG J,WEN J,et al.Locality preserving projection with symmetric graph embedding for unsupervised dimensionality reduction[J].Pattern Recognition,2022,131:108844.
[13] LU Y W,YUAN C,LI X L,et al.Structurally incoherent low-rank 2DLPP for image classification[J].IEEE Transactions on Circuits and Systems for Video Technology,2018,29(6):1701-1714.
[14] SONG B,TIAN Y,NAJLA A N. Simultaneous p-and s-orders minmax robust locality preserving projection[J].Multimedia Tools and Applications,2022,81(29):42513-42526.
[15] CHEN S S,GUO W. Auto-encoders in deep learning:a review with new perspectives[J].Mathematics,2023,11(8):1777.
[16] 刘会芸,侯志平.基于自动编码器数据降维的滚动轴承故障诊断研究[J].现代制造工程,2022(1):148-153,160.
[17] JI Z,PANG Y W,HE Y Q,et al.Semi-supervised LPP algorithms for learning-to-rank-based visual search reranking[J].Information Sciences An International Journal,2015,302:83-93.
[18] RAN R S,FENG J,LI Z,et al.Locality preserving projections with autoencoder[J].Expert Systems with Applications,2024,242:122750.
[19] SMITH W A,RANDALL R B. Rolling element bearing diagnostics using the Case Western Reserve University data:a benchmark study[J].Mechanical Systems and Signal Processing,2015,64:100-131.
[20] 王斐,许波.基于自适应LPP特征降维和改进VPMCD的滚动轴承故障诊断[J].现代制造工程,2024(6):94,154-161.
[21] HE X,CAI D,NIYOGI P. Laplacian score for feature selection[C]//Proceedings of the 18th International Conference on Neural Information Processing Systems. Cambridge:MIT Press,2005:507-514.
[22] KVÅLSETH T O. On normalized mutual information:measure derivations and properties[J].Entropy,2017:19(12):631.
[23] SANTOS J M, EMBRECHTS M. On the use of the adjusted rand index as a metric for evaluating supervised classification[C]//Proceedings of the 19th International Conference on Artificial Neural Networks:Part II. Berlin:Springer,2009:175-184.

基于收缩自编码器和局部保持投影的机械故障特征提取^*

Mechanical fault feature extraction based on contractive auto-encoder and locality preserving projections

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	陈家芳, 唐湛恒, 周健. 基于改进SDP和FasterNet-GCAM的滚动轴承故障诊断[J]. 现代制造工程, 2025, 538(7): 129-138.
[2]	何新荣, 邵峰, 郭嘉, 杜小泽. 变工况下基于IPNCC-MCSKNet的滚动轴承故障声纹识别方法^*[J]. 现代制造工程, 2025, 536(5): 135-143.
[3]	张霞, 梁海波, 高原, 万夫, 李泉昌, 仇芝, 缐傲航. 基于PCA-RF协同的GRU网络滚动轴承退化趋势预测^*[J]. 现代制造工程, 2025, 534(3): 132-140.
[4]	乔宁宁, 李峰. 基于电流信号的变速工况行星齿轮箱故障诊断^*[J]. 现代制造工程, 2025, 532(1): 111-120.
[5]	夏之罡, 楼小波, 马爱军, 翁兴辰. 基于频域格拉姆角场与多尺度深度残差网络的高频变负载滚动轴承故障诊断研究^*[J]. 现代制造工程, 2025, 532(1): 137-147.
[6]	邢海波, 李杰. 基于时频自监督学习的弱标记滚动轴承故障诊断研究^*[J]. 现代制造工程, 2025, 532(1): 148-155.
[7]	董红平, 朱代根, 范辉. 基于改进MSE算法的滑动轴承故障特征提取研究^*[J]. 现代制造工程, 2025, 532(1): 156-162.
[8]	郭康, 王志刚, 徐增丙. 基于AFF-Stablenet模型的小样本轴承故障诊断^*[J]. 现代制造工程, 2024, 528(9): 144-151.
[9]	王斐, 许波. 基于自适应LPP特征降维和改进VPMCD的滚动轴承故障诊断^*[J]. 现代制造工程, 2024, 525(6): 154-161.
[10]	刘满强, 吴杰. 基于ICEEMDAN与POA-SVM的感应电机故障诊断^*[J]. 现代制造工程, 2024, 524(5): 127-137.
[11]	牛礼民, 万凌初, 胡超. 基于IPSO-MCKD的汽车变速箱轴承故障诊断^*[J]. 现代制造工程, 2024, 522(3): 134-139.
[12]	纪京生, 周莉, 马向阳. VMD结合小波包信息熵和GJO-SVM的电机轴承故障诊断[J]. 现代制造工程, 2024, 521(2): 128-136.
[13]	肖伟, 李泽军, 管天福, 贺路, 陈绪兵. 基于PSO-IBP神经网络的纯电动汽车电驱总成故障诊断^*[J]. 现代制造工程, 2024, 520(1): 137-141.
[14]	胡晨;李志敏. 基于多特征阈值判据的航空液压柱塞泵故障诊断方法研究[J]. 现代制造工程, 2023, 516(9): 145-151.
[15]	莫坚;张泽. 轴承剩余使用寿命的注意力多尺度卷积神经网络预测[J]. 现代制造工程, 2023, 515(8): 148-154.