Image feature extraction from wavelet scalogram based on kernel principle component analysis
Article Text (Baidu Translation)
-
摘要: 分析了转子不平衡、不对中、碰摩及油膜涡动的尺度谱图像特征, 提出利用核主成分分析(KPCA)对故障信号的小波尺度谱进行特征提取的方法。利用ZT-3多功能转子试验台获取上述4种故障各32个样本, 对其进行连续小波变换和KPCA特征提取, 并同时提取了相同样本条件下的尺度谱纹理特征和频谱特征。最后利用参数自适应支持向量机模型对提取的特征进行了分类测试。分析结果表明: KPCA方法所提取特征的平均识别效果均达到90%以上, 高于尺度谱纹理特征和频谱特征的分类结果, 能够有效提取尺度谱的特征, 有利于转子故障的智能诊断。Abstract: The scalogram image features of unbalance, misalignment, rub-impact and oil whirl fault were analyzed, and a new feature extraction method from the wavelet scalogram of fault signals was put forward based on kernel principle component analysis(KPCA).By using ZT-3 multi-functional rotor test bed, 32 samples for each type of fault were obtained, continuous wavelet transformation was carried out, and KPCA feature, scalogram texture feature and spectrum feature were extracted.Finally, the extracted features were tested and classified by using parameter self-adaptive support vector machine.Analysis result shows that the average recognition effect of features extracted by KPCA is up to 90%, and is higher than the classification results of scalogram texture feature and spectrum feature, so KPCA can effectively extract the features of scalogram and is helpful for the intelligent diagnosis of rotor faults.
-
Key words:
- wavelet scalogram /
- feature extraction /
- fault diagnosis /
- rotor /
- KPCA
-
表 1 主元贡献率
Table 1. Contribution rates of principle component
特征值序号 p阶多项式核函数(p=1.5) 高斯径向基核函数(σ=105) λi值 贡献率/% 累积贡献率/% λi值 贡献率/% 累积贡献率/% 1 2.075 4 67.28 67.28 4.193 69.75 69.75 2 0.623 1 20.20 87.48 1.052 17.49 87.25 3 0.139 4 4.52 92.00 0.227 3.78 91.03 4 0.058 1 1.88 93.88 0.150 2.49 93.52 5 0.031 5 1.02 94.90 0.077 1.28 94.80 6 0.020 6 0.67 95.57 0.050 0.83 95.62 7 0.013 8 0.45 96.02 0.027 0.45 96.07 8 0.013 5 0.44 96.46 0.024 0.40 96.47 
下载: 导出CSV
表 2 识别结果比较
Table 2. Comparison of recognition results %
特征 测试1 验证1 测试2 验证2 测试3 验证3 测试4 验证4 平均 频谱特征 88.37 82.05 76.74 90.70 79.07 88.37 90.70 81.40 84.68 纹理特征 88.37 90.70 94.29 84.61 90.70 84.62 93.02 83.72 88.75 KPCA特征1 95.35 88.37 88.37 94.29 91.43 94.29 97.67 100.00 93.72 KPCA特征2 94.29 97.67 97.67 100.00 95.35 90.07 95.35 90.70 95.14
下载: 导出CSV
-
[1] PENG Z, HE Y, LU Q, et al. Feature extraction of the rub-i mpact rotor systemby means of wavelet analysis[J]. Journal of Sound and Vibration, 2003, 259(4): 1000-1010. doi: 10.1006/jsvi.2002.5376 [2] 彭志科, 何永勇, 褚福磊. 小波尺度谱在振动信号分析中的应用研究[J]. 机械工程学报, 2002, 38(3): 122-126. https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB200203026.htmPENG Zhi-ke, HE Yong-yong, CHU Fu-lei. Using wavelet scalogramfor vibration signals analysis[J]. Chinese Journal of Mechanical Engineering, 2002, 38(3): 122-126. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB200203026.htm [3] 陈果, 邓堰. 转子故障的连续小波尺度谱特征提取新方法[J]. 航空动力学报, 2009, 24(4): 793-798. https://www.cnki.com.cn/Article/CJFDTOTAL-HKDI200904014.htmCHEN Guo, DENG Yan. New approach of features extraction for rotor faults from continuous wavelet transforms calogram[J]. Journal of Aerospace Power, 2009, 24(4): 793-798. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-HKDI200904014.htm [4] 侯敬宏, 黄树红, 申弢, 等. 基于小波分析的旋转机械振动信号定量特征研究[J]. 机械工程学报, 2004, 40(1): 131-135. https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB200401025.htmHOU Jing-hong, HUANG Shu-hong, SHEN Tao, et al. Wavelet-based quantitative analysis of vibration signal of rotary machines[J]. Chinese Journal of Mechanical Engineering, 2004, 40(1): 131-135. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB200401025.htm [5] SCHOLKOPF B, SMOLA A, MULLER K R. Nonlinear component analysis as a kernel eigenvalue problem[J]. Neural Computation, 1998, 10(5): 1299-1319. doi: 10.1162/089976698300017467 [6] KI M KI, JUNG K, KI M HJ. Face recognition using kernel principal component analysis[J]. IEEE Signal Processing Letters, 2002, 9(2): 40-42. doi: 10.1109/97.991133 [7] MALHI A, GAO R X. PCA-based feature selection scheme for machine defect classification[J]. IEEE Transactions on Instrumentation and Measurement, 2004, 53(6): 1517-1525. [8] HE Qian, LIU Yi-bing, PENG Lu. Kernel principal components analysis for earlyidentification of gear tooth crack[C]∥IEEE. Proceedings of the6th World Congress on Intelligent Control and Automation. Dalian: IEEE, 2006: 5748-5751. [9] 李巍华, 廖广兰, 史铁林. 核函数主元分析及其在齿轮故障诊断中的应用[J]. 机械工程学报, 2003, 39(8): 65-70. https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB200308013.htmLI Wei-hua, LI AO Guang-lan, SHI Tie-lin. Kernel principal component analysis andits applicationin gear fault diagnosis[J]. Chinese Journal of Mechanical Engineering, 2003, 39(8): 65-70. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JXXB200308013.htm [10] 陈果. 基于遗传算法的支持向量机分类器模型参数优化[J]. 机械科学与技术, 2007, 26(3): 347-350. doi: 10.3321/j.issn:1003-8728.2007.03.019CHEN Guo. Opti mizing the parameters of support vector machines classifier model based on genetic algorithm[J]. Mechanical Science and Technology, 2007, 26(3): 347-350. (in Chinese) doi: 10.3321/j.issn:1003-8728.2007.03.019 -
点击查看大图
图(2) / 表(2)
计量
- 文章访问数: 604
- HTML全文浏览量: 184
- PDF下载量: 298
- 被引次数: 0
