A Novel Method of Quality Abnormality Detection and Fault Quantitative Assessment for Industrial Processes
-
摘要: 质量异常检测(Quality abnormality detection, QAD)与故障量化评估(Fault quantitative assessment, FQA)作为工业过程监控的关键环节, 是故障诊断领域的研究热点. 本文提出了一种新的工业过程质量异常检测与故障量化评估方法. 首先, 采用弹性网络(Elastic net, EN)算法构建了质量相关的变量候选集, 借助典型相关分析(Canonical correlation analysis, CCA)构建了质量相关的特征向量, 并引入支持向量数据描述(Support vector data description, SVDD)实现质量异常检测. 其次, 从优化近邻点距离的角度提出了增强局部线性嵌入(Enhanced local linear embedding, ELLE)算法, 并提出了基于CCA-ELLE的质量异常故障量化评估方法. 最后, 通过田纳西−伊斯曼(Tennessee-Eastman, TE)过程进行仿真验证, 并与传统的方法进行对比分析, 实验结果验证了所提方法的优越性和有效性.Abstract: As the key link of industrial process monitoring, quality abnormality detection (QAD) and fault quantitative assessment (FQA) are the research hotspot in the field of fault diagnosis. In this paper, a novel framework of industrial process QAD and FQA is proposed. Firstly, the candidate set of quality related variables is constructed by elastic net (EN) algorithm. The quality related eigenvectors are constructed by canonical correlation analysis (CCA), and the support vector data description (SVDD) is introduced to detect quality anomaly. Secondly, an enhanced local linear embedding (ELLE) algorithm is proposed from the perspective of optimizing the distance between adjacent points, and a quantitative assessment method based on CCA- ELLE is proposed. Finally, the proposed framework is simulated and compared with some traditional methods by Tennessee-Eastman (TE) process simulation platform. The experimental results verify the superiority and effectiveness of the proposed method.
-
表 1 TE过程变量
Table 1 Process variables in the TE process
变量描述 变量描述 1) 物料 A 流量 12) 气/液分离器液位 2) 物料 D 流量 13) 气/液分离器压力 3) 物料 E 流量 14) 气/液分离器出口流量 4) 物料 C 流量 15) 汽提塔液位 5) 压缩机返回物料流量 16) 汽提塔压力 6) 反应器给料流量 17) 汽提塔塔底流量 7) 反应器压力 18) 汽提塔温度 8) 反应器液位 19) 汽提塔蒸汽流量 9) 反应器温度 20) 压缩机功率 10) 排空物料流量 21) 反应器冷却水出口温度 11) 气/液分离器温度 22) 冷凝器冷却水出口温度 表 2 验证数据集
Table 2 Data sets used for validation
故障程度 样本数量 正常状态样本 (TE 标准数据) 500 (训练集) 故障数据 (TE 标准数据) 960 (测试集) 正常状态样本 (生成数据 FS = 0) 500 + 160 (训练集 + 测试集) 故障程度 1 (生成数据 FS = 0.2) 160 (测试集) 故障程度 2 (生成数据 FS = 0.4) 160 (测试集) 故障程度 3 (生成数据 FS = 0.6) 160 (测试集) 故障程度 4 (生成数据 FS = 0.8) 160 (测试集) 故障程度 5 (生成数据 FS = 1.0) 160 (测试集) 表 3 两种方法的性能比较
Table 3 Comparison of the two methods
方法 误报率 (%) 检测率 (%) 建模运行时间 (s) KPLS 3.75 98.375 0.307 CCA-SVDD 0 97.75 0.033 -
[1] 彭开香, 马亮, 张凯. 复杂工业过程质量相关的故障检测与诊断技术综述. 自动化学报, 2017, 43(3): 349-365Peng Kai-Xiang, Ma Liang, Zhang Kai. Review of quality-related fault detection and diagnosis techniques in complex industrial processes. Acta Automatica Sinica, 2017, 43(3): 349-365 [2] Krzanowski W J. A user's guide to principal components. Journal of the Royal Statistical Society: Series A, 1992, 155(1): 173-174 doi: 10.2307/2982678 [3] Qin S J. Survey on data-driven industrial process monitoring and diagnosis. Annual Reviews in Control, 2012, 36(2): 220-234 doi: 10.1016/j.arcontrol.2012.09.004 [4] 马洁, 李钢, 陈默. 基于非线性故障重构的旋转机械故障预测方法. 自动化学报, 2014, 40(9): 2045-2049Ma Jie, Li Gang, Chen Mo. Fault prediction method of rotating machinery based on nonlinear fault reconstruction. Acta Automatica Sinica, 2014, 40(9): 2045-2049 [5] 王静, 胡益, 侍洪波. 基于GMM的间歇过程故障检测. 自动化学报, 2015, 41(5): 899-905Wang Jing, Hu Yi, Shi Hong-Bo. Intermittent process fault detection based on GMM. Acta Automatica Sinica, 2015, 41(5): 899-905 [6] Zhou D H, Li G, Qin S J. Total projection to latent structures for process monitoring. AIChE Journal, 2010, 56(1): 168-178 [7] Ding S X, Yin S, Peng K X, Hao H Y, Shen B. A novel scheme for key performance indicator prediction and diagnosis with application to an industrial hot strip mill. IEEE Transactions on Industrial Informatics, 2013, 9(4): 2239-2247 doi: 10.1109/TII.2012.2214394 [8] Huang J P, Yan X F. Quality-driven principal component analysis combined with kernel least squares for multivariate statistical process monitoring. IEEE Transactions on Control System Technology, 2019, 27(6): 2688-2695 doi: 10.1109/TCST.2018.2865130 [9] Wang G, Yin S. Quality-related fault detection approach based on orthogonal signal correction and modified PLS. IEEE Transactions on Industrial Informatics, 2015, 11(2): 398-405 [10] Jiang Y C, Yin S. Recent advances in key-performance-indicator oriented prognosis and diagnosis with a matlab toolbox: DB-KIT. IEEE Transactions on Industrial Informatics, 2019, 15(5): 2849-2858 doi: 10.1109/TII.2018.2875067 [11] 尚林源, 田学民, 曹玉苹, 蔡连芳. 基于PLS交叉积矩阵非相似度分析的MPC性能监控与诊断. 自动化学报, 2017, 43(2): 271-279Shang Lin-Yuan, Tian Xue-Min, Cao Yu-Ping, Cai Lian-Fang. Performance monitoring and diagnosis of MPC based on PLS cross product matrix non-similarity analysis. Acta Automatica Sinica, 2017, 43(2): 271-279 [12] Chen Z, Ding S X, Zhang K. Canonical correlation analysis-based fault detection methods with application to alumina evaporation process. Control Engineering Practice, 2016, 46: 51-58 doi: 10.1016/j.conengprac.2015.10.006 [13] Zhang K, Peng K X, Ding S X, Chen Z W, Yang X. A correlation-based distributed fault detection method and its application to a hot tandem rolling mill process. IEEE Transactions on Industrial Electronics, 2013, 9(4): 2226-2238 doi: 10.1109/TII.2013.2243743 [14] David M J T, Robert P W D. Support vector domain description. Pattern Recognition Letters, 1999, 20(11): 1191-1199 [15] Khediri I B, Weihs C, Limam M. Kernel k-means clustering based local support vector domain description fault detection of multimodal processes. Expert Systems with Applications, 2012, 39(2): 2166-2171 doi: 10.1016/j.eswa.2011.07.045 [16] Zhang C F, Peng K X, Dong J. A novel plant-wide process monitoring framework based on distributed Gap-SVDD with adaptive radius. Neurocomputing, 2019, 350: 1-12 doi: 10.1016/j.neucom.2019.04.026 [17] Zhu K, Mei F, Zheng J. Adaptive fault diagnosis of HVCBs based on P-SVDD and P-KFCM. Neurocomputing, 2017, 240: 127-136 doi: 10.1016/j.neucom.2017.02.042 [18] Zhu Q, Qin S J. Supervised diagnosis of quality and process faults with canonical correlation analysis. Industrial and Engineering Chemistry Research, 2019, 58(26): 11213-11223 doi: 10.1021/acs.iecr.9b00320 [19] Ma L, Dong J, Peng K X. A novel key performance indicator oriented hierarchical monitoring and propagation path identification framework for complex industrial processes. ISA Transactions, 2020, 96(1): 1-13 [20] Luo L J, Lovelett R J, Ogunnaike B A. Hierarchical monitoring of industrial processes for fault detection, fault grade evaluation, and fault diagnosis. AIChE Journal, 2017, 63(7): 2781-2795 doi: 10.1002/aic.15662 [21] Song B, Zhou X G, Shi H B, Yang T. Performance indicator oriented concurrent subspace process monitoring method. IEEE Transactions on Industrial. Electronics, 2018, 65(2): 1508-1571 doi: 10.1109/TIE.2017.2733443 [22] Yang T, Shi H B, Song B, Tan S. Operating performance assessment and non-optimal cause identification for chemical process. Canadian. Journal of Chemical. Engineering, 2017, 97(1): 1475-1487 [23] Guo S, Sun Y, Wu F, Li Y. Integrating laplacian eigenmaps feature space conversion into deep neural network for equipment condition assessment. Automatic Control and Computer Sciences, 2018, 52(6): 465-475 doi: 10.3103/S0146411618060056 [24] Yan H, Liu K, Zhang X. Multiple sensor data fusion for degradation modeling and prognostics under multiple operational conditions. IEEE Transactions on Reliability, 2016, 65(3): 1416-1426 doi: 10.1109/TR.2016.2575449 [25] Sun C, Wang P, Yan R Q, Gao R X, Chen X F. Machine health monitoring based on locally linear embedding with kernel sparse representation for neighborhood optimization bearing fault diagnosis. Mechanical System and Signal Processing, 2019, 114: 25-34 doi: 10.1016/j.ymssp.2018.04.044 [26] Atamuradov V, Medjaher K, Camci F, Dersin P, Zerhouni N. Railway point machine prognostics based on feature fusion and health state assessment. IEEE Transactions on Instrumentation and Measurement, 2019, 68(8): 2691-2704 doi: 10.1109/TIM.2018.2869193 [27] Zhang F, Sun K, Wu X L. A novel variable selection algorithm for multi-layer perceptron with elastic net. Neurocomputing, 2019, 361: 110-118 doi: 10.1016/j.neucom.2019.04.091 [28] Downs J, Vogel E F. A plant-wide industrial process control problem. Computers and Chemical Engineering, 1993, 17(3): 245-255 doi: 10.1016/0098-1354(93)80018-I