Multi-kernel Multi-class Relevance Vector Machine and its Application to Fault Diagnosis of Rolling Bearing with Multi-feature Fusion
-
摘要: 考虑到滚动轴承振动信号的复杂性, 单一故障特征难以获得较理想的故障诊断结果, 提出一种基于多核多分类相关向量机(Multi-kernel multi-class relevance vector machine, MMRVM)的多特征融合智能故障诊断方法。该方法将具有不同特性的故障特征通过核函数映射到高维特征空间, 按照特征贡献量大小进行加权求和从而融合形成多特征空间, 充分利用各特征向量的有效属性, 有效避免不同特征直接融合导致的维数增高问题。此外, 通过量子遗传算法自适应选取不同特征对应的最优核参数, 进一步提高了故障识别准确率。滚动轴承故障诊断实例表明, 与其它方法相比, 所提方法可有效融合多种滚动轴承故障特征信息, 具有更高的故障诊断准率。Abstract: Considering the complexity of rolling bearing vibration signals, it is difficult to obtain ideal fault diagnosis results only using single fault feature, a novel multi-feature fusion intelligent fault diagnosis method based on multi-kernel multi-class relevance vector machine was proposed. According to the contribution of fault features, different fault features are mapped to high dimensional feature space and fused to multiple feature spaces with weighted summation in this method. It can overcome the curse of dimensionality if different features are integrated directly, because the effective properties of different feature vector are fully utilized. Additionally, corresponding kernel parameters of different fault features are selected adaptively by quantum genetic algorithm, and the fault identification accuracy can be further improved. Compared with other methods, the experiment of rolling bearing fault diagnosis shows that the proposed method can fuse various fault features of rolling bearing effectively and achieve higher fault diagnosis accuracy.
-
Key words:
- fault diagnosis /
- rolling bearing /
- feature fusion /
- multi-kernel multi-class /
- relevance vector machine
-
表 1 各特征向量类别及其对应的表示方法
特征向量类别 基本属性 小波包时域特征向量 [x0, x1, …, xn] EEMD域特征向量 [c0, c1, …, cN] 统计域特征向量 [Xrms2, Ip, Cf, Ce] 表 2 不同特征融合方法的故障诊断性能对比情况
融合特征类别 训练时间/s Rvs/个 准确率/% EEMD 10.01 13 98.08 WPT 9.46 10 97.44 时域 7.36 4 33.46 EEMD+WPT 12.94 9 99.36 WPT+时域 10.45 10 98.95 EEMD+时域 11.77 12 99.23 EEMD+WPT+时域 12.96 7 99.87 EEMD∩WPT∩时域 28.75 16 98.92 注: EEMD∩WPT∩时域为3种特征直接混合 表 3 不同诊断模型的故障诊断实验结果
故障识别器 Rvs或Svs/个 准确率/% QGA-SVM 19 94.23 MMRVM 10 96.37 QGA-MMRVM 7 99.87 KNN - 94.16 BPNN - 94.06 -
[1] 何正嘉, 曹宏瑞, 訾艳阳, 等. 机械设备运行可靠性评估的发展与思考[J]. 机械工程学报, 2014, 50(2): 171-186 https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB201402027.htmHE Z J, CAO H R, ZI Y Y, et al. Developments and thoughts on operational reliability assessment of mechanical equipment[J]. Journal of Mechanical Engineering, 2014, 50(2): 171-186 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB201402027.htm [2] ZHAO D F, ZHANG H L, LIU S L, et al. Deep rational attention network with threshold strategy embedded for mechanical fault diagnosis[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 3519715 [3] ZHAO H Y, WANG J D, LEE J, et al. A compound interpolation envelope local mean decomposition and its application for fault diagnosis of reciprocating compressors[J]. Mechanical Systems and Signal Processing, 2018, 110: 273-295 doi: 10.1016/j.ymssp.2018.03.035 [4] SHAO H D, JIANG H K, ZHANG H Z, et al. Rolling bearing fault feature learning using improved convolutional deep belief network with compressed sensing[J]. Mechanical Systems and Signal Processing, 2018, 100: 743-765 doi: 10.1016/j.ymssp.2017.08.002 [5] 李恒, 张氢, 秦仙蓉, 等. 基于短时傅里叶变换和卷积神经网络的轴承故障诊断方法[J]. 振动与冲击, 2018, 37(19): 124-131 https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201819021.htmLI H, ZHANG Q, QIN X R, et al. Fault diagnosis method for rolling bearings based on short-time Fourier transform and convolution neural network[J]. Journal of Vibration and Shock, 2018, 37(19): 124-131 https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201819021.htm [6] 陈法法, 杨晶晶, 肖文荣, 等. Adaboost-SVM集成模型的滚动轴承早期故障诊断[J]. 机械科学与技术, 2018, 37(2): 237-243 https://www.cnki.com.cn/Article/CJFDTOTAL-JXKX201802013.htmCHEN F F, YANG J J, XIAO W R, et al. Early fault diagnosis of rolling bearing based on ensemble model of Adaboost SVM[J]. Mechanical Science and Technology for Aerospace Engineering, 2018, 37(2): 237-243 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXKX201802013.htm [7] 王波, 刘树林, 蒋超, 等. 基于量子遗传算法优化RVM的滚动轴承智能故障诊断[J]. 振动与冲击, 2015, 34(17): 207-212WANG B, LIU S L, JIANG C, et al. Rolling bearings' intelligent fault diagnosis based on RVM optimized with Quantum genetic algorithm[J]. Journal of Vibration and Shock, 2015, 34(17): 207-212 (in Chinese) [8] YAN X A, LIU Y, JIA M P. A fault diagnosis approach for rolling bearing integrated SGMD, IMSDE and multiclass relevance vector machine[J]. Sensors, 2020, 20(15): 4352 doi: 10.3390/s20154352 [9] ZENG M, YANG Y, LUO S R, et al. One-class classification based on the convex hull for bearing fault detection[J]. Mechanical Systems and Signal Processing, 2016, 81: 274-293 doi: 10.1016/j.ymssp.2016.04.001 [10] ZHAO D F, LIU S L, GU D, et al. Enhanced data-driven fault diagnosis for machines with small and unbalanced data based on variational auto-encoder[J]. Measurement Science and Technology, 2020, 31(3): 035004 doi: 10.1088/1361-6501/ab55f8 [11] PSORAKIS I, DAMOULAS T, GIROLAMI M A. Multiclass relevance vector machines: sparsity and accuracy[J]. IEEE Transactions on Neural Networks, 2010, 21(10): 1588-1598 doi: 10.1109/TNN.2010.2064787 [12] 唐勇波, 丰娟. KTA-SVM的变压器油中溶解气体浓度预测[J]. 控制工程, 2017, 24(11): 2263-2267 https://www.cnki.com.cn/Article/CJFDTOTAL-DWJS202007044.htmTANG Y B, FENG J. A prediction method of dissolved gas content in transformer oil based on KTA-SVM[J]. Control Engineering of China, 2017, 24(11): 2263-2267 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-DWJS202007044.htm [13] SHAPIRO A. Monte Carlo sampling methods[J]. Handbooks in Operations Research and Management Science, 2003, 10(3): 353-425 [14] DAMOULAS T, GIROLAMI M A. Combining feature spaces for classification[J]. Pattern Recognition, 2009, 42(11): 2671-2683 doi: 10.1016/j.patcog.2009.04.002 [15] 杨俊安, 庄镇泉, 史亮. 多宇宙并行量子遗传算法[J]. 电子学报, 2004, 32(6): 923-928 doi: 10.3321/j.issn:0372-2112.2004.06.011YANG J A, ZHUANG Z Q, SHI L. Multi-universe parallel quantum genetic algorithm[J]. Acta Electronica Sinica, 2004, 32(6): 923-928 (in Chinese) doi: 10.3321/j.issn:0372-2112.2004.06.011 [16] WANG B, LIU S L, ZHANG H L, et al. Fault diagnosis of rolling bearing based on relevance vector machine and kernel principal component analysis[J]. Journal of Vibroengineering, 2014, 16(1): 57-69 [17] 陈法法, 汤宝平, 董绍江. 基于粒子群优化LS-WSVM的旋转机械故障诊断[J]. 仪器仪表学报, 2011, 32(12): 2747-2753 https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201112018.htmCHEN F F, TANG B P, DONG S J. Rotating machinery fault diagnosis based on LS-WSVM with particle swarm optimization[J]. Chinese Journal of Scientific Instrument, 2011, 32(12): 2747-2753 (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-YQXB201112018.htm