基于加权深度支持向量数据描述的工业过程故障检测

doi:10.11949/0438-1157.20210707

摘要/Abstract

摘要：

传统支持向量数据描述（SVDD）方法本质上采用浅层学习框架，难以有效监控非线性工业过程的复杂故障。针对此问题，提出一种基于加权深度支持向量数据描述（WDSVDD）的故障检测方法。该方法一方面在深度学习框架下重新定义SVDD优化目标函数，构建基于深度特征的深度SVDD监控模型（DSVDD），并利用核密度估计法计算监控指标的统计控制限；另一方面，考虑到深度特征的故障敏感度差异特性，在DSVDD监控模型中设计特征加权层，分别从静态和动态信息分析角度给出权重因子的计算方法，利用权重因子突出故障敏感特征的影响以提高故障检测率。应用于一个典型化工过程的测试结果表明，所研究的方法能够比传统SVDD方法更有效地监控过程中复杂故障的发生。

关键词: 动态建模, 过程系统, 算法, 故障检测, 深度学习, 支持向量数据描述, 非线性过程, 加权因子

Abstract:

The traditional support vector data description (SVDD) method essentially uses a shallow learning framework, which makes it difficult to effectively monitor complex faults in nonlinear industrial processes. To solve this problem, a fault detection method based on weighted deep support vector data description (WDSVDD) is proposed. On the one hand, the objective function of SVDD optimization is redefined in the framework of deep learning, and a deep SVDD monitoring model (DSVDD) based on deep features is constructed. The kernel density estimation method is used to calculate the statistical control limit of monitoring indicator. On the other hand, considering the fault sensitivity difference of deep features, a feature weighting layer is added in the DSVDD monitoring model. The weighting factors are computed from the perspectives of the static and dynamic information analysis, respectively, which are used to highlight the influence of fault-sensitive features for improving the fault detection rate. The testing results on one typical chemical process show that the proposed method can monitor the occurrence of complex faults more effectively than the traditional SVDD method.

Key words: dynamic modeling, process systems, algorithm, fault detection, deep learning, support vector data description, nonlinear processes, weighting factor

中图分类号:

TP 277

王晓慧, 王延江, 邓晓刚, 张政. 基于加权深度支持向量数据描述的工业过程故障检测[J]. 化工学报, 2021, 72(11): 5707-5716.

Xiaohui WANG, Yanjiang WANG, Xiaogang DENG, Zheng ZHANG. Industrial process fault detection using weighted deep support vector data description[J]. CIESC Journal, 2021, 72(11): 5707-5716.

图/表 11

图1 SVDD模型结构解析

Fig.1 Structure analytic diagram of SVDD model

图2 WDSVDD模型结构

Fig.2 Structure diagram of WDSVDD model

图3 TE过程流程图

Fig.3 Flowchart of TE process

表1 用于算法测试的故障

Table 1 Faults for algorithm testing

编号	故障描述
IDV4	反应器冷却水入口温度发生阶跃变化
IDV5	冷凝器冷却水入口温度发生阶跃变化
IDV10	C进料温度随机波动
IDV11	反应器冷却水入口温度随机波动
IDV16	未知类型故障
IDV17	未知类型故障
IDV19	未知类型故障
IDV20	未知类型故障
IDV21	物流4的阀门故障

图4 不同方法对故障IDV10的监控效果

Fig.4 Monitoring charts for fault IDV10 based on different methods

图5 不同方法对故障IDV19的监控效果

Fig.5 Monitoring charts for fault IDV19 based on different methods

图6 加权前后的深度特征对比分析

Fig.6 Deep features comparison before and after weighting

表2 故障检测率比较

Table 2 Comparison of fault detection rates

No.	FDR/%
No.	SVDD	DSVDD	SWSVDD	DWSVDD
IDV4	95.50	100.00	100.00	100.00
IDV5	37.63	100.00	100.00	100.00
IDV10	59.38	78.63	83.13	95.25
IDV11	72.50	71.88	77.50	97.50
IDV16	46.38	82.38	86.25	98.25
IDV17	91.75	96.75	97.25	97.75
IDV19	16.50	62.50	70.75	95.63
IDV20	62.63	72.75	77.25	92.13
IDV21	47.50	54.88	60.63	65.75
mean	58.86	79.97	83.64	93.58

表3 故障误报率比较

Table 3 Comparison of false fault alarming rates

No.	FAR/%
No.	SVDD	DSVDD	SWSVDD	DWSVDD
IDV4	1.88	0.63	4.38	0.63
IDV5	1.88	0.63	4.38	0.63
IDV10	1.88	5.63	10.00	5.63
IDV11	3.13	3.13	10.00	3.13
IDV16	28.75	16.88	23.13	18.13
IDV17	1.25	6.88	12.5	6.88
IDV19	0.63	4.38	6.25	4.38
IDV20	0.63	1.25	3.13	1.25
IDV21	8.13	7.50	17.50	8.13
mean	5.35	5.21	10.14	5.42

表4 不同调节因子情形下的平均监控效果

Table 4 Average monitoring results under different tuning factors

$γ$	SWDSVDD		DWDSVDD
$γ$	FAR/%	FDR/%	FAR/%	FDR/%
0.5	14.93	86.68	5.28	93.21
1	12.85	85.31	5.35	93.32
2	10.14	83.64	5.42	93.58
4	7.08	81.58	5.63	93.90
8	5.56	80.42	6.04	94.08
16	5.21	79.97	6.39	94.13

表4 不同调节因子情形下的平均监控效果

Table 4 Average monitoring results under different tuning factors

$γ$	SWDSVDD		DWDSVDD
$γ$	FAR/%	FDR/%	FAR/%	FDR/%
0.5	14.93	86.68	5.28	93.21
1	12.85	85.31	5.35	93.32
2	10.14	83.64	5.42	93.58
4	7.08	81.58	5.63	93.90
8	5.56	80.42	6.04	94.08
16	5.21	79.97	6.39	94.13

图7 窗宽对DWDSVDD监控效果的影响

Fig.7 Influence of window width on the DWDSVDD monitoring results

参考文献 30

1	Booyse W, Wilke D N, Heyns S. Deep digital twins for detection, diagnostics and prognostics[J]. Mechanical Systems and Signal Processing, 2020, 140: 106612.
2	李元, 杨东昇, 赵丽颖, 等. 层次变分高斯混合模型与主多项式分析的故障检测策略[J]. 化工学报, 2021, 72(3): 1616-1626.
	Li Y, Yang D S, Zhao L Y, et al. Fault detection using hierarchical variational Gaussian mixture model and principal polynomial analysis[J]. CIESC Journal, 2021, 72(3): 1616-1626.
3	Chen H T, Jiang B, Zhang T Y, et al. Data-driven and deep learning-based detection and diagnosis of incipient faults with application to electrical traction systems[J]. Neurocomputing, 2020, 396: 429-437.
4	Zhu J L, Ge Z Q, Song Z H, et al. Review and big data perspectives on robust data mining approaches for industrial process modeling with outliers and missing data[J]. Annual Reviews in Control, 2018, 46: 107-133.
5	徐静, 王振雷, 王昕. 基于非线性动态全局局部保留投影算法的化工过程故障检测[J]. 化工学报, 2020, 71(12): 5655-5663.
	Xu J, Wang Z L, Wang X. Fault detection for chemical process based on nonlinear dynamic global-local preserving projections[J]. CIESC Journal, 2020, 71(12): 5655-5663.
6	姚羽曼, 罗文嘉, 戴一阳. 数据驱动方法在化工过程故障诊断中的研究进展[J]. 化工进展, 2021, 40(4): 1755-1764.
	Yao Y M, Luo W J, Dai Y Y. Research progress of data-driven methods in fault diagnosis of chemical process[J]. Chemical Industry and Engineering Progress, 2021, 40(4): 1755-1764.
7	Wang B, Mao Z Z. A dynamic ensemble outlier detection model based on an adaptive k-nearest neighbor rule[J]. Information Fusion, 2020, 63: 30-40.
8	Wang K Z, Lan H B. Robust support vector data description for novelty detection with contaminated data[J]. Engineering Applications of Artificial Intelligence, 2020, 91: 103554.
9	Zhao Y P, Xie Y L, Ye Z F. A new dynamic radius SVDD for fault detection of aircraft engine[J]. Engineering Applications of Artificial Intelligence, 2021, 100: 104177.
10	Jiang Q C, Yan X F, Huang B. Deep discriminative representation learning for nonlinear process fault detection[J]. IEEE Transactions on Automation Science and Engineering, 2020, 17(3): 1410-1419.
11	王建林, 马琳钰, 刘伟旻, 等. 基于核相似度支持向量数据描述的间歇过程监测[J]. 化工学报, 2017, 68(9): 3494-3500.
	Wang J L, Ma L Y, Liu W M, et al. Batch process monitoring by kernel similarity-based support vector data description [J]. CIESC Journal, 2017, 68(9): 3494-3500.
12	赵小强, 牟淼. 基于变量分块的KDLV-DWSVDD间歇过程故障检测算法研究[J]. 仪器仪表学报, 2021, 42(2): 244-256.
	Zhao X Q, Mou M. Research on fault detection algorithm of batch process based on KDLV-DWSVDD of variable blocks[J]. Chinese Journal of Scientific Instrument, 2021, 42(2): 244-256.
13	Zhang Y F, Li X S. Two-step support vector data description for dynamic, non-linear, and non-Gaussian processes monitoring[J]. The Canadian Journal of Chemical Engineering, 2020, 98(10): 2109-2124.
14	Yuan P, Mao Z Z, Wang B. A pruned support vector data description-based outlier detection method: applied to robust process monitoring[J]. Transactions of the Institute of Measurement and Control, 2020, 42(11): 2113-2126.
15	Liu C Y, Gryllias K. A semi-supervised Support Vector Data Description-based fault detection method for rolling element bearings based on cyclic spectral analysis[J]. Mechanical Systems and Signal Processing, 2020, 140: 106682.
16	Lv Z, Yan X F, Jiang Q C, et al. Just-in-time learning-multiple subspace support vector data description used for non-Gaussian dynamic batch process monitoring[J]. Journal of Chemometrics, 2019, 33(6): e3134.
17	Haghighi F, Omranpour H. Stacking ensemble model of deep learning and its application to Persian/Arabic handwritten digits recognition[J]. Knowledge-Based Systems, 2021, 220: 106940.
18	Wang M, Deng W H. Deep face recognition: a survey[J]. Neurocomputing, 2021, 429: 215-244.
19	Garain A, Singh P K, Sarkar R. FuzzyGCP: a deep learning architecture for automatic spoken language identification from speech signals[J]. Expert Systems with Applications, 2021, 168: 114416.
20	Tax D M J, Duin R P W. Support vector data description[J]. Machine Learning, 2004, 54(1): 45-66.
21	谢彦红, 贾冬妮, 张成, 等. 基于邻域保持嵌入-支持向量数据描述的过程监控算法及其应用[J]. 信息与控制, 2020, 49(5): 625-632.
	Xie Y H, Jia D N, Zhang C, et al. Support-vector-data-description process monitoring algorithm based on neighborhood preserving embedding and its application[J]. Information and Control, 2020, 49(5): 625-632.
22	Wang X H, Wang Y J, Deng X G, et al. A batch process monitoring method using two-dimensional localized dynamic support vector data description[J]. IEEE Access, 2020, 8: 181192-181204.
23	Ruff L, Vandermeulen R A, Gornitz N, et al. Deep one-class classification [C]//The 35th International Conference on Machine Learning. Stockholm, Sweden, 2018: 4390-4399.
24	Ruff L, Vandermeulen RA, Gornitz N, et al. Deep support vector data description for unsupervised and semi-supervised anomaly detection [C]//The ICML 2019 Workshop on Uncertainty and Robustness in Deep Learning. Long Beach, California, USA, 2019: 1-10.
25	Odiowei P E P, Cao Y. Nonlinear dynamic process monitoring using canonical variate analysis and kernel density estimations[J]. IEEE Transactions on Industrial Informatics, 2010, 6(1): 36-45.
26	Downs J J, Vogel E F. A plant-wide industrial process control problem[J]. Computers & Chemical Engineering, 1993, 17(3): 245-255.
27	Xue T, Zhong M Y, Luo L J, et al. Distributionally robust fault detection by using kernel density estimation[J]. IFAC-PapersOnLine, 2020, 53(2): 652-657.
28	Deng X G, Tian X M, Chen S, et al. Deep principal component analysis based on layerwise feature extraction and its application to nonlinear process monitoring[J]. IEEE Transactions on Control Systems Technology, 2019, 27(6): 2526-2540.
29	Chen Q, Wang Y Q. Key-performance-indicator-related state monitoring based on kernel canonical correlation analysis[J]. Control Engineering Practice, 2021, 107: 104692.
30	Cao J, He Y L, Zhu Q X. An ontology-based procedure knowledge framework for the process industry[J]. The Canadian Journal of Chemical Engineering, 2021, 99(2): 530-542.

[1]	陈哲文, 魏俊杰, 张玉明. 超临界水煤气化耦合SOFC发电系统集成及其能量转化机制[J]. 化工学报, 2023, 74(9): 3888-3902.
[2]	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.
[3]	温凯杰, 郭力, 夏诏杰, 陈建华. 一种耦合CFD与深度学习的气固快速模拟方法[J]. 化工学报, 2023, 74(9): 3775-3785.
[4]	张逸豪, 王振雷. 基于最大信息系数的分组支持向量数据描述故障检测[J]. 化工学报, 2023, 74(9): 3865-3878.
[5]	尹刚, 李伊惠, 何飞, 曹文琦, 王民, 颜非亚, 向禹, 卢剑, 罗斌, 卢润廷. 基于KPCA和SVM的铝电解槽漏槽事故预警方法[J]. 化工学报, 2023, 74(8): 3419-3428.
[6]	郑玉圆, 葛志伟, 韩翔宇, 王亮, 陈海生. 中高温钙基材料热化学储热的研究进展与展望[J]. 化工学报, 2023, 74(8): 3171-3192.
[7]	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.
[8]	李贵贤, 曹阿波, 孟文亮, 王东亮, 杨勇, 周怀荣. 耦合固体氧化物电解槽的CO₂制甲醇过程设计与评价研究[J]. 化工学报, 2023, 74(7): 2999-3009.
[9]	周继鹏, 何文军, 李涛. 异形催化剂上乙烯催化氧化失活动力学反应工程计算[J]. 化工学报, 2023, 74(6): 2416-2426.
[10]	邵远哲, 赵忠盖, 刘飞. 基于共同趋势模型的非平稳过程质量相关故障检测方法[J]. 化工学报, 2023, 74(6): 2522-2537.
[11]	高学金, 姚玉卓, 韩华云, 齐咏生. 基于注意力动态卷积自编码器的发酵过程故障监测[J]. 化工学报, 2023, 74(6): 2503-2521.
[12]	邵伟明, 韩文学, 宋伟, 杨勇, 陈灿, 赵东亚. 基于分布式贝叶斯隐马尔可夫回归的动态软测量建模方法[J]. 化工学报, 2023, 74(6): 2495-2502.
[13]	贠程, 王倩琳, 陈锋, 张鑫, 窦站, 颜廷俊. 基于社团结构的化工过程风险演化路径深度挖掘[J]. 化工学报, 2023, 74(4): 1639-1650.
[14]	宋冰, 郑城风, 侍洪波, 陶阳, 谭帅. 基于VAE-OCCA的质量相关故障检测方法研究[J]. 化工学报, 2023, 74(4): 1630-1638.
[15]	郑书闽, 郭鹏程, 颜建国, 王帅, 李文博, 周淇. 微小通道内过冷流动沸腾阻力特性实验及预测研究[J]. 化工学报, 2023, 74(4): 1549-1560.