Improved Relative-transformation Principal Component Analysis Based on Mahalanobis Distance and Its Application for Fault Detection
-
摘要: 目前, 主元分析方法(PCA)在数据处理、模式识别、过程监测等领域得到了越来越广泛的应用, 但仍存在部分关键问题亟待解决. 本文为了提高PCA方法的故障检测性能, 进行了一系列的改进, 首先, 本文引入相对变换的概念, 使用马氏距离相对变换直接消除量纲, 通过理论推导证明了马氏距离相对变换可以对数据不进行标准化直接进行数据变换, 而且给出了在相对空间内数据进行PCA变换的合理解释, 表明了基于马氏距离相对变换的PCA故障检测方法可以有效的消除变量量纲对数据的影响, 提高数据的可分性. 其次, 改进了SPE监控指标, 提出一种基于马氏距离的平方预测误差指标, 更有效地实现对工业过程的故障检测. 最后, 将两种改进方法相结合, 提出改进的马氏距离相对变换PCA故障检测方法, 并以轧钢过程活套系统为背景, 实际数据仿真结果表明: 与PCA以及其它改进方法相比, 本文提出的方法具有更好的故障检测性能和实时性, 能准确、有效地检测出活套故障.Abstract: Principal component analysis (PCA) has been widely used in process industries, which could maintain the maximum fault detection rate. Although many issues have been addressed in PCA, some essential problems remain unresolved. This study improves PCA for fault detection performance in the following ways. Firstly, a relative transformation scheme based on Mahalanobis distance (MD) is introduced to eliminate the effect of dimension of data instead of dimensionless standardization, and improve the accuracy and real-time performance of fault detection. The theoretical derivation proves that relative transformation based on MD can directly eliminate the effect of dimension and give reasonable explanation of PCA in the relative space, the analysis and simulation results show its superiority and effectiveness. Secondly, an improved squared prediction error (SPE) statistic is given to improve the fault detection performance of standardized PCA, which can make the standardized PCA-based fault detection method more suitable for the actual industrial process. Finally, two improved methods are combined to detect the fault more effectively. The proposed methods are applied to detect single fault and multi-fault of looper system in hot continuous rolling process, simulation results demonstrate the effectiveness of these improvements for fault detection performance in terms of sensitiveness, accuracy and real-time performance of fault detection.
-
[1] Garcia-Alvarez D, Fuente M J, Sainz G I. Fault detection and isolation in transient states using principal component analysis. Journal of Process Control, 2012, 22(3): 551-563 [2] Shi Huai-Tao, Liu Jian-Chang, Ding Xiao-Di, Tan Shuai, Wang Xue-Mei. Fault detection based on hybrid dynamic principal component analysis. Control Engineering of China, 2012, 19(1): 148-151 (in Chinese) [3] Wen Cheng-Lin, Hu Yu-Cheng. Fault diagnosis based on information incremental matrix. Acta Automatica Sinica, 2012, 38(5): 832-840 (in Chinese) [4] Wu Feng, Zhong Yan, Wu Quan-Yuan. Online classification framework for data stream based on incremental kernel principal component analysis. Acta Automatica Sinica, 2010, 36(4): 534-542 (in Chinese) [5] Dobos L, Abonyi J. On-line detection of homogeneous operation ranges by dynamic principal component analysis based time-series segmentation. Chemical Engineering Science, 2012, 75: 96-105 [6] Zhang Y W, Zhou H, Qin S J. Decentralized fault diagnosis of large-scale processes using multiblock kernel principal component analysis. Acta Automatica Sinica, 2010, 36(4): 593-597 [7] Luan Fang-Jun, Guo Hong-Mei, Lin Lan, Wang Yong-Hui. Algorithm of face authentication based on wavelet transform and combining two different 2-dimensional participial component analyses. Journal of Shenyang Jianzhu University (Natural Science), 2010, 26(5): 1001-1005 (in Chinese) [8] Chiang L H, Russell E L, Braatz R D. Fault Detection and Diagnosis in Industrial Systems. Berlin: Springer, 2001 [9] Wen Cheng-Lin, Hu Jing, Wang Tian-Zhen, Chen Zhi-Guo. Relative PCA with applications of data compression and fault diagnosis. Acta Automatica Sinica, 2008, 34(9): 1128-1139 (in Chinese) [10] Dunia R, Qin S J, Edgar T F, McAvoy T J. Identification of faulty sensors using principal component analysis. American Institute of Chemical Engineers Journal, 1996, 42(10): 2797-2812 [11] Zhao Zhong-Gai, Liu Fei. Application research of statistical monitoring index based on Mahalanobis distance. Acta Automatica Sinica, 2008, 34(4): 493-495 (in Chinese) [12] Wen G H, Jiang L J, Wen J. Local relative transformation with application to isometric embedding. Pattern Recognition Letters, 2009, 30(3): 203-211 [13] Li Yuan, Tang Xiao-Chu. Improved performance of fault detection based on selection of the optimal number of principal components. Acta Automatica Sinica, 2009, 35(2): 1550-1557 (in Chinese) [14] Tamura M, Tsujita S. A study on the number of principal components and sensitivity of fault detection using PCA. Computers and Chemical Engineering, 2007, 31(9): 1035-1046 [15] Banerjee A, Burlina P. Efficient particle filtering via sparse kernel density estimation. IEEE Transactions on Image Processing, 2010, 19(9): 2480-2490 [16] Shi H T, Liu J C, Zhang Y, Li L. Fault detection method based on relative-transformation partial least squares. Chinese Journal of Scientific Instrument, 2012, 33(4): 816-822 [17] Marcu T, Köppen-Seliger B, Stücher R. Design of fault detection for a hydraulic looper using dynamic neural networks. Control Engineering Practice, 2008, 16(2): 192-213 [18] Yue Yu-Mei, Ma Ya-Li, Wang Jiao-Qing. Research about fault intelligent diagnosis of high speed rotating machinery. Journal of Shenyang Jianzhu University (Natural Science), 2005, 21(6): 770-773 (in Chinese)
点击查看大图
计量
- 文章访问数: 1491
- HTML全文浏览量: 71
- PDF下载量: 1355
- 被引次数: 0