基于KECA的非线性工业过程故障检测与诊断新方法

doi:10.11949/0438-1157.20191518

摘要/Abstract

摘要：

提出了一种基于核熵成分分析（kernel entropy component analysis，KECA）的非线性过程故障检测与诊断新方法。该方法首先利用KECA获取过程数据的得分向量及非线性特征子空间；然后鉴于KECA可以以角结构的方式揭示数据中潜在的集群结构，设计了基于角度的监测指标VoA。该指标通过各得分向量之间的角度方差来描述变换后数据间的结构差异,并根据角度方差的变化情况实现故障检测；接着，为了在检测到故障后有效地进行故障识别，构建了KECA相似度因子来度量特征子空间的相似程度以识别故障模式；最后，以非线性数值案例及Tennessee Eastman过程进行仿真测试研究，结果验证了所提方法的可行性及有效性。

关键词: 故障检测与诊断, 核熵成分分析, VoA监测指标, 过程控制, 相似度因子, 模型, 安全

Abstract:

A new method for non-linear process fault detection and diagnosis based on kernel entropy component analysis (KECA) is proposed. Firstly, the score vectors and nonlinear feature space are obtained using KECA. Since KECA can reveal the underlying cluster structure in the data in the form of the angular structure, an angle-based monitoring index referred to as vector of angle (VoA) is designed. This index measures the structural difference between the transformed data by the angle variance of each score vector, and realizes fault detection according to the change. Then, in order to effectively identify faults after detection, KECA similarity factor is constructed to measure the similarity of feature space to identify fault patterns. Finally, the feasibility and validity of the proposed method are demonstrated by the nonlinear numerical case and Tennessee Eastman (TE) process.

Key words: fault detection and diagnosis, kernel entropy component analysis, VoA monitoring index, process control, similarity factors, model, safety

中图分类号:

TP 277

邓明月, 刘建昌, 许鹏, 谭树彬, 商亮亮. 基于KECA的非线性工业过程故障检测与诊断新方法[J]. 化工学报, 2020, 71(5): 2151-2163.

Mingyue DENG, Jianchang LIU, Peng XU, Shubin TAN, Liangliang SHANG. New fault detection and diagnosis strategy for nonlinear industrial process based on KECA[J]. CIESC Journal, 2020, 71(5): 2151-2163.

图/表 14

图1 非线性数据集及其KECA变换

Fig.1 Nonlinear data set and KECA transformation

图2 基于角度的离群点检测示意图

Fig.2 Schematic diagram of angle-based outlier detection

图3 基于KECA的过程监测流程图

Fig.3 Flowchart of process monitoring based on KECA

图4 基于KECA的故障识别流程

Fig.4 Flowchart of fault identification based on KECA

图5 微小故障1的检测结果

Fig.5 Fault detection results of minor fault 1

图6 微小故障2的检测结果

Fig.6 Fault detection results of minor fault 2

表1 两种微小故障的检测率

Table 1 Fault detection rate of two minor faults/%

故障

类型

KECA

KPCA

VoA

T²

98.75

图7 故障4的检测结果

Fig.7 Fault detection results of fault 4

图8 故障16的检测结果

Fig.8 Fault detection results of fault 16

图9 故障21的检测结果

Fig.9 Fault detection results of fault 21

表2 TE过程21种故障的故障检测率、误报率及检测延迟

Table 2 Fault detection rate, false alarm rate and detection latency of 21 faults in TE process

故障序号	FDR/%			FAR/%			DL/num
	KECA		KPCA	KECA		KPCA	KECA		KPCA
	T²	VoA	T²	T²	VoA	T²	T²	VoA	T²
1	99.75	99.88	99.13	0	0	0	2	1	7
2	98.13	99.25	98.38	0	0	0	15	14	14
4	100	100	58.88	0	0	0.63	0	0	2
5	24.63	27.13	24.75	0	0	0.63	0	0	0
6	100	100	99.13	0	0	0.63	0	0	8
7	100	100	100	0	0	0	0	0	0
8	97.38	97.75	97	0	0	1.25	19	18	25
10	69.13	73.5	30.38	0	0	0	26	26	57
11	74.25	77.85	55.25	0	1.25	0	5	5	5
12	99.13	99.38	98.63	3.75	1.25	1.25	2	2	2
13	95.13	95.25	93.63	0	0	0	37	37	48
14	100	100	99.88	0	0	0.63	0	0	0
16	85.85	87.88	14	6.88	2.5	2.5	6	6	297
17	92.88	94.13	80.25	0	0	0.63	23	21	28
18	89.63	89.75	89.38	0	0	1.25	84	84	87
19	15.75	24.38	13.75	0	0	0	10	10	10
20	47.38	53	37.88	0	0	0.63	87	84	84
21	42.5	50	42	0	0	1.25	494	449	505
平均值	79.53	81.55	68.46	0.59	0.28	0.62	45	42	65

表3 用于故障识别的TE过程数据集

Table 3 TE process datasets for fault identification

数据集	描述	样本数
F01H～F21H	故障1～故障21的历史故障模式数据集	480
F01E～F21E	故障1～故障21的早期数据集	200
F01T～F21T	故障1～故障21的测试数据集	800

图10 故障7的F07E与F01H～F21H之间的KECASF

Fig.10 KECASF between F07E of fault 7 and fault mode data set F01H—F21H

表4 TE过程故障测试数据集的识别结果

Table 4 Identification results of TE process fault test data set

故障测试数据集	$N A L L$	$N P C A S F$	$N K P C A S F$	$N K E C A S F$	$η P C A S F$ /%	$η K P C A S F$ /%	$η K E C A S F$ /%
F01T	15	15	15	15	100	100	100
F02T	15	15	15	15	100	100	100
F03T	15	0	2	8	0	13.3	53.3
F04T	15	15	15	15	100	100	100
F05T	15	2	3	13	13.3	20	86.7
F06T	15	14	9	13	93.33	60	86.7
F07T	15	13	15	14	86.7	100	93.3
F08T	15	8	8	13	53.3	53.3	86.7
F09T	15	3	4	7	20	26.7	46.7
F10T	15	6	5	8	40	33.3	53.3
F11T	15	14	12	12	93.3	80	80
F12T	15	9	11	15	60	73.3	100
F13T	15	4	2	8	26.7	13.3	53.3
F14T	15	15	15	15	100	100	100
F15T	15	0	5	9	0	33.3	60
F16T	15	6	9	10	40	60	66.7
F17T	15	15	15	14	100	100	93.3
F18T	15	11	12	11	73.3	80	73.3
F19T	15	13	14	7	86.7	93.3	46.7
F20T	15	15	15	15	100	100	100
F21T	15	12	13	11	80	86.7	73.3
平均值	15	9.8	10.2	11.8	65.1	67.9	78.7

表4 TE过程故障测试数据集的识别结果

Table 4 Identification results of TE process fault test data set

故障测试数据集	$N A L L$	$N P C A S F$	$N K P C A S F$	$N K E C A S F$	$η P C A S F$ /%	$η K P C A S F$ /%	$η K E C A S F$ /%
F01T	15	15	15	15	100	100	100
F02T	15	15	15	15	100	100	100
F03T	15	0	2	8	0	13.3	53.3
F04T	15	15	15	15	100	100	100
F05T	15	2	3	13	13.3	20	86.7
F06T	15	14	9	13	93.33	60	86.7
F07T	15	13	15	14	86.7	100	93.3
F08T	15	8	8	13	53.3	53.3	86.7
F09T	15	3	4	7	20	26.7	46.7
F10T	15	6	5	8	40	33.3	53.3
F11T	15	14	12	12	93.3	80	80
F12T	15	9	11	15	60	73.3	100
F13T	15	4	2	8	26.7	13.3	53.3
F14T	15	15	15	15	100	100	100
F15T	15	0	5	9	0	33.3	60
F16T	15	6	9	10	40	60	66.7
F17T	15	15	15	14	100	100	93.3
F18T	15	11	12	11	73.3	80	73.3
F19T	15	13	14	7	86.7	93.3	46.7
F20T	15	15	15	15	100	100	100
F21T	15	12	13	11	80	86.7	73.3
平均值	15	9.8	10.2	11.8	65.1	67.9	78.7

参考文献 31

1	Yin S, Ding S X, Haghani A, et al. A comparison study of basic data-driven fault diagnosis and process monitoring methods on the benchmark Tennessee Eastman process[J]. Journal of Process Control, 2012, 22(9): 1567-1581.
2	Ge Z Q, Song Z H, Gao F R. Review of recent research on data-based process monitoring[J]. Industrial & Engineering Chemistry Research, 2013, 52(10): 3543-3562.
3	刘强, 卓洁, 郎自强, 等. 数据驱动的工业过程运行监控与自优化研究展望[J]. 自动化学报, 2018, 44(11): 26-38.
	Liu Q, Zhuo J, Lang Z Q, et al. Perspectives on data-driven operation monitoring and self-optimization of industrial processes[J]. Acta Automatica Sinica, 2018, 44(11): 26-38.
4	赵帅, 宋冰, 侍洪波. 基于加权互信息主元分析算法的质量相关故障检测[J]. 化工学报, 2018, 69(3): 962-973.
	Zhao S, Song B, Shi H B. Quality-related fault detection based on weighted mutual information principal component analysis[J]. CIESC Journal, 2018, 69(3): 962-973.
5	于飞, 王红蛟. 基于主元分析与偏最小二乘故障诊断算法的研究[J]. 化工自动化及仪表, 2014, 41(8): 881-883.
	Yu F, Wang H J. Fault diagnosis algorithm research based on principal component analysis and partial least squares[J]. Control and Instruments in Chemical Industry, 2014, 41(8): 881-883.
6	Ge Z Q, Xie L, Kruger U, et al. Local ICA for multivariate statistical fault diagnosis in systems with unknown signal and error distributions[J]. AIChE Journal, 2012, 58(8): 2357-2366.
7	Deng X G, Tian X M, Chen S, et al. Nonlinear process fault diagnosis based on serial principal component analysis[J]. IEEE Transactions on Neural Networks and Learning Systems, 2018, 29(3): 560-572.
8	蔡连芳, 田学民, 张妮. 一种基于改进KICA的非高斯过程故障检测方法[J]. 化工学报, 2012, 63(9): 2864-2868.
	Cai L F, Tian X M, Zhang N. Non-Gaussian process fault detection method based on modified KICA[J]. CIESC Journal, 2012, 63(9): 2864-2868.
9	Wang H, Yao M. Fault detection of batch processes based on multivariate functional kernel principal component analysis[J]. Chemometrics and Intelligent Laboratory Systems, 2015, 149(9): 78-89.
10	Zhao C H, Huang B. Incipient fault detection for complex industrial processes with stationary and nonstationary hybrid characteristics[J]. Industrial & Engineering Chemistry Research, 2018, 57(14): 5045-5057.
11	Jenssen R. Kernel entropy component analysis[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(5): 847-860.
12	Chen J Y, Yu J. Independent component analysis mixture model based dissimilarity method for performance monitoring of non-Gaussian dynamic processes with shifting operating conditions[J]. Industrial & Engineering Chemistry Research, 2014, 53(13): 5055-5066.
13	Gomez-Chova L, Jenssen R, Camps-Valls G. Kernel entropy component analysis for remote sensing image clustering[J]. IEEE Geoscience and Remote Sensing Letters, 2012, 9(2): 312-316.
14	Shekar B H, Kumari M S, Mestetskiy L M, et al. Face recognition using kernel entropy component analysis[J]. Neurocomputing, 2011, 74(6): 1053-1057.
15	Li Y Y, Wang Y, Jiao L, et al. Quantum clustering using kernel entropy component analysis[J]. Neurocomputing, 2016, 202(3): 36-48.
16	Jiang Q C, Yan X F, Lv Z M, et al. Fault detection in nonlinear chemical processes based on kernel entropy component analysis and angular structure[J]. Korean Journal of Chemical Engineering, 2013, 30(6): 1181-1186.
17	Yang Y H, Li X L, Liu X Z, et al. Wavelet kernel entropy component analysis with application to industrial process monitoring[J]. Neurocomputing, 2015, 147(5): 395-402.
18	齐咏生, 张海利, 高学金, 等. 基于KECA的化工过程故障监测新方法[J]. 化工学报, 2016, 67(3): 1063-1069.
	Qi Y S, Zhang H L, Gao X J, et al. Novel fault monitoring strategy for chemical process based on KECA[J]. CIESC Journal, 2016, 67(3): 1063-1069.
19	Miller P, Swanson R E, Heckler C E. Contribution plots: a missing link in multivariate quality control[J]. Appl. Math. Comput. Sci., 1998, 8(4): 775-792.
20	Macgregor J F, Kourti T. Statistical process control of multivariate processes[J]. Control Engineering Practice, 1995, 3(3): 403-414.
21	Karpenko M, Sepehri N, Scuse D. Diagnosis of process valve actuator faults using a multilayer neural network[J]. Control Engineering Practice, 2003, 11(11): 1289-1299.
22	Chiang L H, Kotanchek M E, Kordon A K. Fault diagnosis based on Fisher discriminant analysis and support vector machines[J]. Computers & Chemical Engineering, 2004, 28(8): 1389-1401.
23	Johannesmeyer M C, Singhal A, Seborg D E. Pattern matching in historical data[J]. AIChE Journal, 2002, 48(9):2022-2038.
24	Singhal A, Seborg D. Pattern matching in historical batch data using PCA[J]. IEEE Control Systems Magazine, 2002, 22(5): 53-63.
25	Deng X G, Tian X M. Nonlinear process fault pattern recognition using statistics kernel PCA similarity factor[J]. Neurocomputing, 2013, 121: 298-308.
26	Kriegel H P, Schubert M, Zimek A. Angle-based outlier detection in high-dimensional data[C]// Proc. 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Las Vegas, Nevada, USA: ACM, 2008: 444-452.
27	Shao J D, Rong G, Lee J M. Learning a data-dependent kernel function for KPCA-based nonlinear process monitoring[J]. Chemical Engineering Research & Design, 2009, 87(11):1471-1480.
28	Lee J M, Yoo C K, Choi S W, et al. Nonlinear process monitoring using kernel principal component analysis[J]. Chemical Engineering Science, 2004, 59(1): 223-234.
29	Downs J J, Vogel E F. A plant-wide industrial process control problem[J]. Computers Chem. Engng., 1993, 17(3): 245-255.
30	Lyman P R, Georgakis C. Plant-wide control of the Tennessee Eastman problem[J]. Computers and Chemical Engineering, 1995, 19(3):321-331.
31	顾炳斌, 熊伟丽. 基于多块信息提取的PCA故障诊断方法[J]. 化工学报, 2019, 70(2): 316-329.
	Gu B B, Xiong W L. Fault diagnosis based on PCA method with multi-block information extraction[J]. CIESC Journal, 2019, 70(2): 316-329.

[1]	宋嘉豪, 王文. 斯特林发动机与高温热管耦合运行特性研究[J]. 化工学报, 2023, 74(S1): 287-294.
[2]	连梦雅, 谈莹莹, 王林, 陈枫, 曹艺飞. 地下水预热新风一体化热泵空调系统制热性能研究[J]. 化工学报, 2023, 74(S1): 311-319.
[3]	金正浩, 封立杰, 李舒宏. 氨水溶液交叉型再吸收式热泵的能量及分析[J]. 化工学报, 2023, 74(S1): 53-63.
[4]	康飞, 吕伟光, 巨锋, 孙峙. 废锂离子电池放电路径与评价研究[J]. 化工学报, 2023, 74(9): 3903-3911.
[5]	李科, 文键, 忻碧平. 耦合蒸气冷却屏的真空多层绝热结构对液氢储罐自增压过程的影响机制研究[J]. 化工学报, 2023, 74(9): 3786-3796.
[6]	王浩, 王振雷. 基于自适应谱方法的裂解炉烧焦模型化简策略[J]. 化工学报, 2023, 74(9): 3855-3864.
[7]	曹跃, 余冲, 李智, 杨明磊. 工业数据驱动的加氢裂化装置多工况切换过渡状态检测[J]. 化工学报, 2023, 74(9): 3841-3854.
[8]	李珍宝, 李超, 王虎, 王绍瑞, 黎泉. MPP抑制铝镁合金粉尘爆炸微观机理研究[J]. 化工学报, 2023, 74(8): 3608-3614.
[9]	李锦潼, 邱顺, 孙文寿. 煤浆法烟气脱硫中草酸和紫外线强化煤砷浸出过程[J]. 化工学报, 2023, 74(8): 3522-3532.
[10]	于旭东, 李琪, 陈念粗, 杜理, 任思颖, 曾英. 三元体系KCl + CaCl₂ + H₂O 298.2、323.2及348.2 K相平衡研究及计算[J]. 化工学报, 2023, 74(8): 3256-3265.
[11]	诸程瑛, 王振雷. 基于改进深度强化学习的乙烯裂解炉操作优化[J]. 化工学报, 2023, 74(8): 3429-3437.
[12]	闫琳琦, 王振雷. 基于STA-BiLSTM-LightGBM组合模型的多步预测软测量建模[J]. 化工学报, 2023, 74(8): 3407-3418.
[13]	郭雨莹, 敬加强, 黄婉妮, 张平, 孙杰, 朱宇, 冯君炫, 陆洪江. 稠油管道水润滑减阻及压降预测模型修正[J]. 化工学报, 2023, 74(7): 2898-2907.
[14]	刘晓洋, 喻健良, 侯玉洁, 闫兴清, 张振华, 吕先舒. 螺旋微通道对掺氢甲烷爆轰传播的影响[J]. 化工学报, 2023, 74(7): 3139-3148.
[15]	刘春雨, 周桓宇, 马跃, 岳长涛. CaO调质含油污泥干燥特性及数学模型[J]. 化工学报, 2023, 74(7): 3018-3027.