基于深度学习的复杂化工过程软测量模型研究与应用

doi:10.11949/j.issn.0438-1157.20181352

化工学报 ›› 2019, Vol. 70 ›› Issue (2): 564-571.DOI: 10.11949/j.issn.0438-1157.20181352

基于深度学习的复杂化工过程软测量模型研究与应用

耿志强^1,²(),徐猛^1,²,朱群雄^1,²,韩永明^1,²(),顾祥柏^1,^2,³

1. 北京化工大学信息科学与技术学院，北京 100029
2. 智能过程系统工程教育部工程研究中心，北京 100029
3. 中石化炼化工程(集团)股份有限公司，北京 100029

收稿日期:2018-11-16 修回日期:2018-11-26 出版日期:2019-02-05 发布日期:2019-02-05
通讯作者: 韩永明
作者简介:<named-content content-type="corresp-name">耿志强</named-content>（1973—），男，博士，教授，<email>gengzhiqiang@mail.buct.edu.cn</email>|韩永明（1987—），男，博士，副教授，<email>hanym@mail.buct.edu.cn</email>
基金资助:
国家自然科学基金项目(61533003, 61603025, 71572008);中央高校基本科研业务费专项资金(XK1802-4)

Research and application of soft measurement model for complex chemical processes based on deep learning

Zhiqiang GENG^1,²(),Meng XU^1,²,Qunxiong ZHU^1,²,Yongming HAN^1,²(),Xiangbai GU^1,^2,³

1. College of Information Science and Technology, Beijing University of Chemical Technology, Beijing 100029, China
2. Engineering Research Center of Intelligent PSE, Ministry of Education, Beijing 100029, China
3. Sinopec Engineering Group Co., Ltd., Beijing 100029, China

Received:2018-11-16 Revised:2018-11-26 Online:2019-02-05 Published:2019-02-05
Contact: Yongming HAN

摘要/Abstract

摘要：

针对复杂化工生产过程中的一些原材料消耗量难以直接测量的问题，提出了一种基于深度学习的软测量方法。该方法基于一段时间的历史数据，利用平稳小波变换提取历史数据中的多尺度信息，然后与每一个时间点的可观测数据进行合并得到完整的数据集，再划分出训练集和测试集，用带有注意力机制的深度学习算法进行训练和泛化，进而建立软测量模型。最后将提出的方法应用到对苯二甲酸（PTA）生产装置乙酸消耗的软测量中。通过与极限学习机 (extreme learning machine，ELM)、多层感知器 (multi-layer perceptron，MLP)以及普通长短期记忆网络(long short-term memory，LSTM)方法比较，结果表明，该模型的预测准确度较高，具有一定的有效性和适用性，同时对PTA生产装置的乙酸消耗量进行预测分析，从而提高产能和降低能耗。

关键词: 长短期记忆网络, 神经网络, 平稳小波变换, 算法, 多尺度, 对苯二甲酸

Abstract:

Because some raw material consumption in complex chemical production process is difficult to measure directly, a soft sensor method based on the deep learning is proposed. Based on a period of the historical data, the proposed method extracts multi-scale information from the historical data using stationary wavelet decomposition. Then the observable data at every point of time are combined to get a complete dataset which is divided into the training dataset and the testing dataset. Moreover, the soft sensor model is trained and obtained by using the depth learning algorithm based on the attention mechanism. Finally, the proposed method is applied to the soft measurement of acetic acid consumption in a terephthalic acid (PTA) production unit. Compared with the extreme learning machine (ELM), multi-layer perceptron (MLP) and common long short-term memory (LSTM) method, the result analysis shows that the validity and the applicability of the proposed model is verified. Meanwhile, the consumption of acetic acid in the PTA production plant is predicted and analyzed to improve the production capacity and reduce the energy consumption.

Key words: long short-term memory, neural networks, stationary wavelet decomposition, algorithm, multiscale, terephthalic acid

中图分类号:

TP 29

耿志强, 徐猛, 朱群雄, 韩永明, 顾祥柏. 基于深度学习的复杂化工过程软测量模型研究与应用[J]. 化工学报, 2019, 70(2): 564-571.

Zhiqiang GENG, Meng XU, Qunxiong ZHU, Yongming HAN, Xiangbai GU. Research and application of soft measurement model for complex chemical processes based on deep learning[J]. CIESC Journal, 2019, 70(2): 564-571.

图/表 9

图1 RNN网络结构

Fig.1 RNN network structure

图2 LSTM结构单元内部模型

Fig.2 Internal model of structural unit of LSTM

图3 PTA溶剂系统装置

Fig.3 PTA solvent system plant

图4 软测量模型流程

Fig.4 Flow chart of soft measurement model

表1 网络模型参数

Table 1 Parameters of Network model

参数名称	数值
输入维度	[None, 10, 36]
时间步幅	10
FC1节点数目	50
LSTM节点数目	50
LSTM层数	1
FC2节点数目	100
各激活函数	$t a n h$
误差函数	MSE
训练数据(组)	93
测试数据(组)	70

表1 网络模型参数

Table 1 Parameters of Network model

参数名称	数值
输入维度	[None, 10, 36]
时间步幅	10
FC1节点数目	50
LSTM节点数目	50
LSTM层数	1
FC2节点数目	100
各激活函数	$t a n h$
误差函数	MSE
训练数据(组)	93
测试数据(组)	70

表2 网络训练结果

Table 2 Training results of proposed model

项目	RMSE	ARGE
训练	0.071	0.119
测试	0.166	0.182

表3 PTA软测量模型对比结果

Table 3 Contrast results for soft measurement in PTA process

项目	MLP	ELM	普通LSTM	本文方法
训练	0.44	0.34	0.268	0.119
测试	0.53	0.49	0.282	0.182

图5 模型运行结果对比

Fig.5 Comparison of running results

图6 模型运行结果分析

Fig.6 Analysis of running results

参考文献 33

1	汪银宏. 我国精对苯二甲酸(PTA)市场分析[J]. 聚酯工业, 2016, 29(5): 5-7.
	WangY H. Analysis of PTA market in China[J]. Polyester Industry, 2016, 29(5): 5-7.
2	崔培崇, 王学雷, 马增良. 基于数据包络分析的乙烯能效分析系统[J]. 计算机与应用化学, 2010, 27(9): 1182-1186.
	CuiP C, WangX L, MaZ L. An energy efficiency analysis system for ethylene production based on DEA[J]. Computers and Applied Chemistry, 2010, 27(9): 1182-1186.
3	HanY, GengZ, GuX, et al. Energy efficiency analysis based on DEA integrated ISM[J]. Engineering Applications of Artificial Intelligence, 2015, 45(C): 80-89.
4	朱群雄, 陈希, 贺彦林,等. 基于PCA-DEA的乙烯装置能效分析[J]. 化工学报, 2015, 66(1): 278-283.
	ZhuQ X, ChenX, HeY L, et al. Energy efficiency analysis for ethylene plant based on PCA-DEA[J]. CIESC Journal, 2015, 66(1): 278-283.
5	DongG S, RenW. A real-time fault monitoring system for the dynamic chemical process based on PCA[J]. Computers & Applied Chemistry, 2010, 27(10): 1357-1360.
6	GengZ Q, HanY M, ZhangY, et al. Data fusion-based extraction method of energy consumption index for the ethylene industry[C]//Life System Modeling and Intelligent Computing, International Conference on Life System Modeling and Simulation, LSMS 2010, and International Conference on Intelligent Computing for Sustainable Energy and Environment, ICSEE 2010. Wuxi, China, 2010: 84-92.
7	HanY M, GengZ Q, QuY X, et al. Linear optimization fusion model based on fuzzy C-means: case study of energy efficiency evaluation in ethylene product plants[J]. Journal of Analytical and Applied Pyrolysis, 2017, 125: 347-355.
8	HuangG B, ZhuQ Y, SiewC K. Extreme learning machine: a new learning scheme of feedforward neural networks[J]. Proc. Int. Joint Conf. Neural Netw., 2004, 2: 985-990.
9	裘日辉. 基于极限学习机的复杂工业过程故障识别研究及应用[D]. 杭州: 浙江大学, 2016.
	QiuR H. Fault identification for complex industrial processes based on extreme learning machine[D]. Hangzhou：Zhejiang University, 2016.
10	李诗瑶, 周良, 刘虎. 基于深度极限学习机的危险源识别算法HIELM[J]. 计算机科学, 2017, 44(5): 89-94.
	LiS Y, ZhouL, LiuH. Hazard identification algorithm based on deep extreme learning machine[J]. Computer Science, 2017, 44(5): 89-94.
11	SipahiS, TimorM. The analytic hierarchy process and analytic network process: an overview of application[J]. Management Decision, 2010, 48(5): 775-808.
12	GengZ, DongJ, ChenJ, et al. A new self-organizing extreme learning machine soft sensor model and its applications in complicated chemical processes[J]. Engineering Applications of Artificial Intelligence, 2017, 62(C): 38-50.
13	PengD, XuY, WangY. Soft-sensing in complex chemical process based on a sample clustering extreme learning machine model[J]. IFAC PapersOnLine, 2015, 48(8): 801-806.
14	贺彦林, 王晓, 朱群雄. 基于主成分分析-改进的极限学习机方法的精对苯二甲酸醋酸含量软测量[J]. 控制理论与应用, 2015, 32(1): 80-85.
	HeY L, WangX, ZhuQ X. Modeling of acetic acid content in purified terephthalic acid solvent column using principal component analysis based improved extreme learning machine[J]. Control Theory & Applications, 2015, 32(1): 80-85.
15	GengZ Q, QinL, HanY M, et al. Energy saving and prediction modeling of petrochemical industries: a novel ELM based on FAHP[J]. Energy, 2017, 122: 350-362
16	GengZ, ChenJ, HanY. Energy efficiency prediction based on PCA-FRBF model: a case study of ethylene industries[J]. IEEE Transactions on Systems Man & Cybernetics Systems, 2017, 47(8):1763-1773.
17	蓝艇, 童楚东, 史旭华. 变量加权型主元分析算法及其在故障检测中的应用[J]. 化工学报, 2017, 68(8): 3177-3182.
	LanT, TongC D, ShiX H. Variable weighted principal component analysis algorithm and its application in fault detection[J]. CIESC Journal, 2017, 68(8): 3177-3182.
18	朱群雄, 石晓赟, 顾祥柏, 等. 基于时序数据融合的乙烯装置能效价值研究及应用[J]. 化工学报, 2010, 61(10): 2620-2626.
	ZhuQ X, ShiX Y, GuX B, et al. Time-series data fusion and its application to energy efficiency value for ethylene plants[J]. CIESC Journal, 2010, 61(10): 2620-2626.
19	HeY L, XuY, GengZ Q, et al. Soft sensor of chemical processes with large numbers of input parameters using auto-associative hierarchical neural network[J]. Chinese Journal of Chemical Engineering, 2015, 23(1): 138-145.
20	HanY M, GengZ Q, ZhuQ X. Energy optimization and prediction of complex petrochemical industries using an improved artificial neural network approach integrating data envelopment analysis[J]. Energy Conversion & Management, 2016, 124: 73-83.
21	JahediG, ArdehaliM M. Wavelet based artificial neural network applied for energy efficiency enhancement of decoupled HVAC system[J]. Energy Conversion & Management, 2012, 54(1): 47-56.
22	TaghavifarH, MardaniA. Applying a supervised ANN (artificial neural network) approach to the prognostication of driven wheel energy efficiency indices[J]. Energy, 2014, 68(8): 651-657.
23	TeichT, RoesslerF, KretzD, et al. Design of a prototype neural network for smart homes and energy efficiency[J]. Procedia Engineering, 2014, 69(1): 603-608.
24	GersF, SchmidhuberJ, CumminsF. Learning to forget: continual prediction with LSTM[J]. Neural Computation, 2000, 12(10): 2451-2471.
25	WangY, HuangM, ZhuX, et al. Attention-based LSTM for aspect-level sentiment classification[C]// Conference on Empirical Methods in Natural Language Processing. 2016:606-615.
26	张冲. 基于Attention-Based LSTM模型的文本分类技术的研究[D]. 南京: 南京大学, 2016.
	ZhangC. Text classification based on attention-based LSTM model[D]. Nanjing: Nanjing University, 2016.
27	XuK, XieL, YaoK. Investigating LSTM for punctuation prediction[C]// International Symposium on Chinese Spoken Language Processing. IEEE, 2017: 1-5.
28	TsinghuaW K, HuangD, YangF, et al. Soft sensor development and applications based on LSTM in deep neural networks[C]//Computational Intelligence (SSCI), 2017 IEEE Symposium Series on. IEEE. 2017: 1-6.
29	ZhaoH, SunS, JinB. Sequential fault diagnosis based on LSTM neural network[J]. IEEE Access, 2018, 6: 12929-12939.
30	ErtlP, LewisR, MartinE, et al. In silico generation of novel, drug-like chemical matter using the LSTM neural network[J]. arXiv preprint arXiv:1712.07449, 2017.
31	欧芬兰, 郭金泉, 钟剑锋,等. 基于平稳小波变换的悬臂梁微小缺陷检测方法研究[J]. 机电工程, 2015, 32(6): 762-767.
	OuF L, GuoJ Q, ZhongJ F, et al. Small crack identification in cantilever beams by stationary wavelet transform[J]. Journal of Mechanical ＆ Electrical Engineering, 2015, 32(6): 762-767.
32	范小龙, 谢维成, 蒋文波, 等. 一种平稳小波变换改进阈值函数的电能质量扰动信号去噪方法[J]. 电工技术学报, 2016, 31(14):219-226.
	FanX L, XieW C, JiangW B, et al. An improved threshold function method for power quality disturbance signal de-noising based on stationary wavelet transform[J]. Transactions of China Electrotechnical Society, 2016, 31(14): 219-226.
33	房贻广, 颜普, 刘武, 等. 基于平稳小波变换和亮度序的局部特征描述子[J]. 计算机工程, 2017,(11): 272-276,280.
	FangY G, YanP, LiuW, et al. Local feature descriptor based on stationary wavelet transform and intensity order[J]. Computer Engineering, 2017,(11): 272-276,280.

[1]	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.
[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]	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.
[4]	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.
[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]	史昊鹏, 钟达文, 廉学新, 张君峰. 朝下多尺度沟槽翅片结构表面沸腾换热实验研究[J]. 化工学报, 2023, 74(7): 2880-2888.
[7]	徐野, 黄文君, 米俊芃, 申川川, 金建祥. 多源信息融合的离心式压缩机喘振诊断方法[J]. 化工学报, 2023, 74(7): 2979-2987.
[8]	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.
[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]	李艳辉, 丁邵明, 白周央, 张一楠, 于智红, 邢利梅, 高鹏飞, 王永贞. 非常规服役超临界锅炉的微纳尺度腐蚀动力学模型建立及应用[J]. 化工学报, 2023, 74(6): 2436-2446.
[11]	黄磊, 孔令学, 白进, 李怀柱, 郭振兴, 白宗庆, 李平, 李文. 油页岩添加对准东高钠煤灰熔融行为影响的研究[J]. 化工学报, 2023, 74(5): 2123-2135.
[12]	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.
[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]	吴选军, 王超, 曹子健, 蔡卫权. 数据与物理信息混合驱动的固定床吸附穿透深度学习模型[J]. 化工学报, 2023, 74(3): 1145-1160.
[15]	张生安, 刘桂莲. 高效太阳能电解水制氢系统及其性能的多目标优化[J]. 化工学报, 2023, 74(3): 1260-1274.

基于深度学习的复杂化工过程软测量模型研究与应用

Research and application of soft measurement model for complex chemical processes based on deep learning

RichHTML

PDF (PC)

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 33

相关文章 15

编辑推荐

Metrics

本文评价