基于mini-1D-CNN模型的TE过程故障诊断

doi:10.16265/j.cnki.issn1003-3033.2023.02.0017

中国安全科学学报 ›› 2023, Vol. 33 ›› Issue (2): 173-178.doi: 10.16265/j.cnki.issn1003-3033.2023.02.0017

基于mini-1D-CNN模型的TE过程故障诊断

杨余¹(), 杨鑫², 王英¹, 翟持³, 张浩¹^,^**()

¹ 西南大学化学化工学院,重庆 400715
² 重庆理工大学化学化工学院,重庆 400054
³ 昆明理工大学化学工程学院,昆明云南 650500

收稿日期:2022-09-22 修回日期:2022-12-15 出版日期:2023-02-28 发布日期:2023-08-28
通讯作者: **张浩(1986—),男,河北平山人,博士,副教授,主要从事过程系统工程、智能化工、时空大数据等方面的研究。E-mail: haozhang@swu.edu.cn。
作者简介:
杨余 (1998—),男,四川岳池人,硕士研究生,研究方向为过程系统工程。E-mail: 305528356@qq.com。
杨鑫，副教授
翟持，讲师
基金资助:
国家自然基金资助(21806131)

Research on TE process fault diagnosis based on mini-1D-CNN model

YANG Yu¹(), YANG Xin², WANG Ying¹, ZHAI Chi³, ZHANG Hao¹^,^**()

¹ School of Chemistry and Chemical Engineering, Southwest University, Chongqing 400715, China
² School of Chemistry and Chemical Engineering, Chongqing University of Technology, Chongqing 400054, China
³ Faculty of Chemical Engineering, Kunming University of Science of Technology, Kunming Yunnan 650500, China

Received:2022-09-22 Revised:2022-12-15 Online:2023-02-28 Published:2023-08-28

摘要/Abstract

摘要：

为提升石化企业过程监测与故障诊断系统性能,满足化工过程故障诊断实时性、时效性的要求,提出一种基于过程历史数据驱动的最小一维卷积神经网络(mini-1D-CNN)的故障诊断模型。首先,通过一维卷积核学习和识别不同故障类型的数据特征,自动提取优势特征并进行故障分类;其次,通过逐步向后回归选择重要特征参数,优化模型结构。利用可实时获取的31个过程变量与操作参数,输入一维卷积神经网络(1D-CNN),监测与诊断田纳西-伊斯曼(TE)过程的主要故障。结果表明:相对于其他故障诊断模型,mini-1D-CNN模型在测试集上故障诊断率(FDR)较高,可达到96.50 %;同时,mini-1D-CNN模型关注于TE过程故障诊断的重要特征参数,在降低参数量及降低训练和测试时间上具有显著优势。

关键词: 最小一维卷积神经网络(mini-1D-CNN), 田纳西-伊斯曼(TE)过程, 故障诊断, 过程监测, 贡献系数

Abstract:

To improve the performance of process monitoring and fault diagnosis system in petrochemical enterprises and meet the requirements of real-time and timeliness of chemical process fault diagnosis, a fault diagnosis model based on mini-1D-CNN driven by process historical data was proposed. First, the data features of different fault types were learned and identified by one-dimensional convolutional kernel, and the dominant features were automatically extracted and classified second, the important feature parameters were selected by stepwise backward regression to optimize the model structure. In this paper, 31 process variables and operating parameters that can be obtained in real time were input into one-dimensional convolutional neural network (1D-CNN) to monitor and diagnose the main faults of TE process. The results show that compared with other fault diagnosis models, the mini-1D-CNN model has a higher fault diagnosis rate(FDR) on test set, which can reach 96.5; at the same time, the mini-1D-CNN model focuses on the important characteristic parameters of TE process fault diagnosis, and has significant advantages in reducing the number of parameters and reducing training and test time.

Key words: minimal-one-dimensional convolutional neural network (mini-1D-CNN ), Tennessee-Eastman(TE) process, fault diagnosis, process monitoring, contribution coefficient

杨余, 杨鑫, 王英, 翟持, 张浩. 基于mini-1D-CNN模型的TE过程故障诊断[J]. 中国安全科学学报, 2023, 33(2): 173-178.

YANG Yu, YANG Xin, WANG Ying, ZHAI Chi, ZHANG Hao. Research on TE process fault diagnosis based on mini-1D-CNN model[J]. China Safety Science Journal, 2023, 33(2): 173-178.

图/表 7

图1

图2

图3

表1

表2

表3

图4

参考文献 19

[1]	万婧, 容振乾, 赵远飞, 等. 2010—2019年我国化工行业较大以上生产安全事故统计分析及启示[J]. 工业安全与环保, 2021, 47(5): 59-63.
	WAN Jing, RONG Zhenqian, ZHAO Yuanfei, et al. Statistical analysis and lessons of major production safety accidentsin chemical industry of China from 2010 to 2019[J]. Industrial Safety and Environmental Protection, 2021, 47(5): 59-63.
[2]	王海顺, 许铭, 辛盼盼, 等. 我国生产安全事故经济损失统计制度改革建议[J]. 中国安全科学学报, 2019, 29(10): 141-146. doi: 10.16265/j.cnki.issn1003-3033.2019.10.022
	WANG Haishun, XU Ming, XIN Panpan, et al. Recommendations for reform on statistical systems of economic losses caused by work safety accidents in China[J]. China Safety Science Journal, 2019, 29(10): 141-146. doi: 10.16265/j.cnki.issn1003-3033.2019.10.022
[3]	王斌, 施祖建, 匡蕾. 基于改进KFDA和RW v-SVM的化工生产系统故障快速诊断[J]. 中国安全科学学报, 2013, 23(8): 84-89.
	WANG Bin, SHI Zujian, KUANG Lei. A fast method for diagnosing fault in chemical production system based on improved KFDA and RW v-SVM[J]. China Safety Science Journal, 2013, 23(8): 84-89.
[4]	TAO Caixia, WANG Xu, GAO Fengyang, et al. Fault diagnosis of photovoltaic array based on deep belief network optimized by genetic algorithm[J]. Chinese Journal of Electrical Engineering, 2020, 6(3): 106-114.
[5]	ZHANG Zhanpeng, ZHAO Jinsong. A deep belief network based fault diagnosis model for complex chemical processes[J]. Computers & Chemical Engineering, 2017, 107: 395-407. doi: 10.1016/j.compchemeng.2017.02.041
[6]	WU Hao, ZHAO Jinsong. Deep convolutional neural network model based chemical process fault diagnosis[J]. Computers & Chemical Engineering, 2018, 115: 185-197. doi: 10.1016/j.compchemeng.2018.04.009
[7]	YU Wanke, ZHAO Chunhui. Broad convolutional neural network based industrial process fault diagnosis with incremental learning capability[J]. IEEE Transactions on Industrial Electronics, 2019, 67(6): 5081-5091. doi: 10.1109/TIE.41
[8]	WANG Yalin, PAN Zhuofu, YUAN Xiaofeng, et al. A novel deep learning based fault diagnosis approach for chemical process with extended deep belief network[J]. ISA Transactions, 2020, 96: 457-467. doi: S0019-0578(19)30290-3 pmid: 31324340
[9]	邓露, 张阳, 戴一阳, 等. 一种基于动态卷积神经网络的故障诊断方法[C]. 2020年过程系统工程年会, 2020: 1-7.
	DENG Lu, ZHANG Yang, DAI Yiyang, et al. A fault diagnosis method based on dynamic convolutional neural network[C]. PSE 2020, 2020: 1-7.
[10]	XU Li, REN J S, LIU Ce, et al. Deep convolutional neural network for image deconvolution[J]. Advances in Neural Information Processing Systems, 2014, 27: 1790-1798.
[11]	LECUN Y, BENGIO Y. The handbook of brain theory and neural networks[M]. MA Arbib: MIT Press, 1995:255-258.
[12]	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90. doi: 10.1145/3065386
[13]	KALCHBRENNER N, GREFENSTETTE E, BLUNSOM P. A convolutional neural network for modelling sentences[C]. 52^nd Annual Meeting of the Association for Computational Linguistics, 2014: 655-665.
[14]	HASAN M J, SOHAIB M, KIM J-M. 1D CNN-based transfer learning model for bearing fault diagnosis under variable working conditions[C]. International Conference on Computational Intelligence in Information System, 2018: 13-23.
[15]	CHEN Shumei, YU Jianbo, WANG Shijin. One-dimensional convolutional neural network-based active feature extraction for fault detection and diagnosis of industrial processes and its understanding via visualization[J]. ISA Transactions, 2022, 122: 424-443. doi: 10.1016/j.isatra.2021.04.042
[16]	BATHELT A, RICKER N L, JELALI M. Revision of the tennessee eastman process model[J]. IFAC-Papers On Line, 2015, 48(8): 309-314.
[17]	卢春红. 基于数据驱动的故障检测与诊断技术及其应用研究[D]. 无锡: 江南大学, 2015.
	LU Chunhong. Research on data-driven fault detection and diagnosis techniques and their applications[D]. Wuxi: Jiangnan University, 2015.
[18]	OKADA S, OHZEKI M, TAGUCHI S. Efficient partition of integer optimization problems with one-hot encoding[J]. Scientific Reports, 2019, 9(1): 1-12. doi: 10.1038/s41598-018-37186-2
[19]	ZHENG Shaodong, ZHAO Jinsong. A new unsupervised data mining method based on the stacked autoencoder for chemical process fault diagnosis[J]. Computers & Chemical Engineering, 2020, 135(6): DOI:10.1016/j.compchemeng.2020.106755. doi: 10.1016/j.compchemeng.2020.106755

状态类别	训练集		测试集
状态类别	FDR	FPR	FDR	FPR
正常	0.989	0.054	0.969	0.020
故障1	1.000	0.000	0.999	0.000
故障2	0.999	0.000	0.981	0.000
故障4	1.000	0.001	0.996	0.001
故障5	1.000	0.000	0.998	0.000
故障7	1.000	0.000	1.000	0.000
故障8	0.988	0.001	0.981	0.001
故障10	0.981	0.002	0.941	0.001
故障11	0.984	0.003	0.917	0.003
故障12	0.991	0.000	0.974	0.001
故障13	0.970	0.003	0.934	0.002
故障14	0.995	0.002	0.942	0.003
故障17	0.959	0.003	0.925	0.003
故障18	0.956	0.002	0.941	0.002
故障19	0.999	0.000	0.995	0.000
故障20	0.963	0.003	0.941	0.001
平均值	0.986	0.005	0.965	0.002

状态类别	DCNN (50)	DCNN (31)	1D-CNN (50)	mini-1D- CNN (31)
正常	0.808	0.890	0.946	0.969
故障1	0.988	1.000	0.996	0.999
故障2	1.000	0.976	0.987	0.981
故障4	1.000	1.000	0.999	0.996
故障5	1.000	1.000	0.996	0.998
故障7	1.000	1.000	1.000	1.000
故障8	0.940	0.910	0.981	0.981
故障10	0.875	0.917	0.941	0.941
故障11	1.000	1.000	0.916	0.917
故障12	0.989	0.989	0.978	0.974
故障13	0.844	0.870	0.952	0.934
故障14	0.938	0.963	0.935	0.942
故障17	0.945	0.973	0.913	0.925
故障18	0.921	0.899	0.946	0.941
故障19	0.962	0.987	0.996	0.995
故障20	0.896	0.922	0.937	0.941
平均值	0.947	0.958	0.960	0.965

排名	特征参数	训练集准确率增长量	测试集准确率增长量
1	产品分离器温度	0.094 5	0.120 9
2	汽提器温度	0.080 5	0.074 4
3	反应器冷却水出口温度	0.074 1	0.067 4
4	反应器冷却水流量	0.039 5	0.066 9
5	反应器压力	0.077 5	0.066 2
6	排放速度(流9)	0.070 7	0.044 8
7	总进料量(流4)	0.039 6	0.041 4
8	产品分离器压力	0.038 9	0.023 1
9	分离器冷却水出口温度	0.008 3	0.022 5
10	反应器温度	0.023 7	0.021 8

基于mini-1D-CNN模型的TE过程故障诊断

Research on TE process fault diagnosis based on mini-1D-CNN model

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 19

相关文章 15

编辑推荐

Metrics

本文评价

[1]	李刚, 卢佩玲, 杨勇. ZPW-2000轨道电路故障诊断算法应用研究[J]. 中国安全科学学报, 2022, 32(6): 53-59.
[2]	林海香, 卢冉, 陆人杰, 李新琴, 赵正祥, 白万胜. 融合BiLSTM-CBA组合模型的高铁车载设备故障诊断[J]. 中国安全科学学报, 2022, 32(6): 79-86.
[3]	余伟, 王志鹏, 刘硕, 何刚. 多级精细模型下的电机匝间短路故障诊断[J]. 中国安全科学学报, 2022, 32(6): 95-102.
[4]	邢玉龙, 王剑, 上官伟, 彭聪, 朱林富. 面向海量不平衡数据的轨道电路故障诊断方法[J]. 中国安全科学学报, 2022, 32(5): 112-118.
[5]	赵丹, 沈志远, 刘晓青. 基于OCISVM的矿井通风系统在线故障诊断[J]. 中国安全科学学报, 2022, 32(10): 76-82.
[6]	古莹奎, 刘平, 林忠海, 邱光琦. 基于Kurtogram与DSCN的滚动轴承故障诊断方法[J]. 中国安全科学学报, 2021, 31(6): 99-105.
[7]	武诣霖 , 陈晓明, 周锋, 金伟峰, 俞嫒妍, 王正佳. 多跨单层网壳施工平台分析与安全监测[J]. 中国安全科学学报, 2021, 31(4): 119-124.
[8]	陈志煌, 刘国恒, 王莹莹, 朱春丽, 单荐, 翟小东. 基于动态贝叶斯网络的水下连接器故障诊断[J]. 中国安全科学学报, 2020, 30(5): 81-87.
[9]	余敬芝, 燕飞, 牛儒. 基于Petri网的形式化安全分析方法^*[J]. 中国安全科学学报, 2019, 29(S2): 138-143.
[10]	王斌，施祖建，匡蕾. 基于改进KFDA和RW v-SVM的化工生产系统故障快速诊断[J]. 中国安全科学学报, 2013, 23(8): 84-.
[11]	宋志雄，朱红岗，李星星. 基于案例推理的大型养路机械故障诊断系统[J]. 中国安全科学学报, 2012, 22(12): 41-.
[12]	曹惠玲，赵明，曲春刚，薛鹏. 可靠性数据在民机故障诊断中的应用研究[J]. 中国安全科学学报, 2011, 21(10): 95-.
[13]	马成正，王洪德. 基于FMEA与RBF神经网络的LPG汽车罐车储罐系统故障诊断[J]. 中国安全科学学报, 2011, 21(1): 99-.
[14]	梁文，陶振文，黄超，马勇，刘显明，付萍萍，杨济海. 光纤电力通信系统故障诊断方法及应用实践[J]. 中国安全科学学报, 2008, 18(4): 104-.
[15]	岳文辉，肖兴明，唐果宁. 图像识别技术及其在机械零件无损检测中的应用[J]. 中国安全科学学报, 2007, 17(3): 156-.