基于BO-LSTM的低温甲醇洗净化气CO2含量预测

doi:10.16085/j.issn.1000-6613.2024-0244

摘要/Abstract

摘要：

低温甲醇洗工艺作为我国煤化工生产中较为成熟的净化工艺，对其进行建模研究有利于工厂智能化建设与发展。本文基于某石化企业低温甲醇洗装置智能化建设的需要，建立了实时数据预处理规则，并基于贝叶斯优化（BO）的长短时记忆（LSTM）神经网络建立净化气CO₂含量预测模型。结果表明，通过人工筛选及校正和最大信息系数（MIC）方法对装置的实时数据预处理后，可将采集的实时数据变量数由84个降低到22个，降低了数据冗余度；运用BO算法对LSTM模型的6个超参数进行调优，根据优化后的超参数组合建立BO-LSTM预测模型，得到BO-LSTM预测模型的评价指标均方根误差（RMSE）、平均绝对误差（MAE）、回归系数（R²）分别为0.0395、0.0275、0.9843，相比于传统的反向传播（BP）与LSTM模型精度更高、回归效果更好，证明了该模型在低温甲醇洗净化气组成预测运用中的可行性与准确性，能够为净化气产品的生产优化做指导。基于低温甲醇洗净化气CO₂含量预测模型的产品优化建模方法可为相关工艺数字化和智能化建设提供思路。

关键词: 低温甲醇洗, 贝叶斯优化, 神经网络, 计算机模拟, 预测

Abstract:

Rectisol process is a relatively mature purification process in China's coal chemical production, and its modeling research is conducive to the intelligent construction and development of the plant. In this paper, based on the need of intelligent construction of rectisol plant in a petrochemical enterprise, real-time data preprocessing rules were established, and the CO₂ content prediction model of purified gas was built based on the Bayesian optimization (BO) long short-term memory (LSTM) neural network. The results showed that after preprocessing the real-time data of the device by manual screening and correction and the maximum information coefficient (MIC) method, the number of variables of the collected real-time data could be reduced from 84 to 22, which reduced the redundancy of the data. The six hyper-parameters of the LSTM model were tuned by using the BO, and the BO-LSTM prediction model was established based on the optimized hyper-parameter combinations. The evaluation indexes of RMSE, MAE and R² of the prediction model were 0.0395, 0.0275 and 0.9843, respectively, which were higher in precision and better in regression effect than the traditional BP and LSTM model, proving the feasibility and accuracy of the model in the prediction of the rectisol purified gas composition, and it could guide the optimization of the production of purified gas products. The product optimization modeling method based on the CO₂ content prediction model of rectisol purified gas could provide ideas for the digital and intelligent construction of related processes.

Key words: rectisol, Bayesian optimization, neural network, computer simulation, prediction

中图分类号:

TQ015

孙悦鹏, 孙延吉, 潘艳秋, 王成宇. 基于BO-LSTM的低温甲醇洗净化气CO₂含量预测[J]. 化工进展, 2025, 44(2): 688-697.

SUN Yuepeng, SUN Yanji, PAN Yanqiu, WANG Chengyu. Prediction of CO₂ content in Rectisol purified gas based on BO-LSTM[J]. Chemical Industry and Engineering Progress, 2025, 44(2): 688-697.

图/表 12

图1 低温甲醇洗工艺流程简图V101—水分离器；T101—甲醇洗涤塔；T102—CO2产品塔；T103—H2S浓缩塔；T104—CO2气提塔；T105—热再生塔；T106—甲醇/水分离塔；T107—尾气洗涤塔；V102—H2S分离器；P101—贫甲醇泵

表1 低温甲醇洗装置实时数据采集表

时间	FIA1001 /m³·h^-1	TIA1001 /℃	PI1002 /MPa	TI1008 /℃	…	AIA1006 /%
11/16/8:00	343.332	26.691	5.323	-17.368	…	3.275
11/16/7:59	340.978	26.553	5.322	-17.423	…	3.278
11/16/7:58	341.077	26.722	5.323	-17.395	…	3.274
$⋮$	$⋮$	$⋮$	$⋮$	$⋮$	…	$⋮$
10/16/8:02	367.376	27.265	5.365	-17.726	…	3.334
10/16/8:01	368.837	27.490	5.364	-17.561	…	3.327
10/16/8:00	364.665	27.425	5.364	-17.539	…	3.295

表1 低温甲醇洗装置实时数据采集表

时间	FIA1001 /m³·h^-1	TIA1001 /℃	PI1002 /MPa	TI1008 /℃	…	AIA1006 /%
11/16/8:00	343.332	26.691	5.323	-17.368	…	3.275
11/16/7:59	340.978	26.553	5.322	-17.423	…	3.278
11/16/7:58	341.077	26.722	5.323	-17.395	…	3.274
$⋮$	$⋮$	$⋮$	$⋮$	$⋮$	…	$⋮$
10/16/8:02	367.376	27.265	5.365	-17.726	…	3.334
10/16/8:01	368.837	27.490	5.364	-17.561	…	3.327
10/16/8:00	364.665	27.425	5.364	-17.539	…	3.295

图2 实时数据预处理规则流程

表2 输入变量与输出变量相关性分析结果示例

变量序号	MIC数值
1	0.1776
2	0.1965
3	0.1575
4	0.2387
5	0.2136
$⋮$	$⋮$
71	0.1831

表2 输入变量与输出变量相关性分析结果示例

变量序号	MIC数值
1	0.1776
2	0.1965
3	0.1575
4	0.2387
5	0.2136
$⋮$	$⋮$
71	0.1831

表3 输入变量位号及含义

序号	位号	含义
1	PI1002	原料气压力
2	TI1003	原料气进入水分离器温度
3	TI1072	尾气洗涤塔塔顶出口物料温度
4	TIA1099	尾气洗涤塔塔底进料温度
5	FICA1005	甲醇洗涤塔中段回流流量
6	TI1010	甲醇洗涤塔中段回流温度
7	TI1013	H₂S浓缩塔中段物料压缩后温度
8	TIA1006	甲醇洗涤塔塔顶出口物料温度
9	TIA1015	甲醇洗涤塔塔顶甲醇进料温度
10	FICA1051	CO₂产品塔塔顶溶液进料流量
11	TI1033	循环甲醇中段换热器出口温度
12	TI1036	H₂S浓缩塔塔底换热器出口温度
13	TI1037	贫甲醇泵出口换热器温度
14	TI1041	CO₂汽提塔塔顶出口物料温度
15	TI1094	H₂S浓缩塔塔顶富甲醇进料温度
16	TIA1031	H₂S浓缩塔中段物料采出温度
17	TIA1076	贫甲醇泵进料温度
18	TIA1106	CO₂产品塔塔顶出口物料温度
19	FIA1038	克劳斯气流量
20	TIA1051	热再生塔底再沸器回流温度
21	TIA1059	克劳斯气温度

图3 LSTM神经网络结构示意图

图4 LSTM隐含层中细胞层结构示意图

图5 BO-LSTM预测模型流程图

表4 预测模型超参数选取范围与优化结果

超参数	取值范围	优化结果
LSTM层1神经元个数	[40,100]	84
LSTM层2神经元个数	[10,50]	21
Dropout层1丢弃概率	[0.1,0.5]	0.4262
Dropout层2丢弃概率	[0.1,0.5]	0.1520
初始学习率	[0.001,0.1]	0.0564
批次大小	[1,50]	38

图6 BO-LSTM模型预测结果R2

表5 神经网络样本模型超参数组合

超参数	模型1	模型2	模型3	模型4	模型5
LSTM层1神经元个数	54	68	97	92	74
LSTM层2神经元个数	46	39	20	18	53
Dropout层1丢弃概率	0.1043	0.4660	0.1222	0.3935	0.4971
Dropout层2丢弃概率	0.1279	0.4137	0.2510	0.2021	0.3734
初始学习率	0.0013	0.0195	0.0208	0.0596	0.0846
批次大小	25	11	14	27	49

图7 模型预测结果评价指标对比

参考文献 29

1	GATTI Manuele, MARTELLI Emanuele, MARECHAL François, et al. Review, modeling, Heat Integration, and improved schemes of Rectisol®‍-based processes for CO₂ capture[J]. Applied Thermal Engineering, 2014, 70(2): 1123-1140.
2	马震伟. 低温甲醇洗技术及其在煤化工中的应用探讨[J]. 中国石油和化工标准与质量, 2023, 43(17): 145-147.
	MA Zhenwei. Discussion on low temperature methanol washing technology and its application in coal chemical industry[J]. China Petroleum and Chemical Standard and Quality, 2023, 43(17): 145-147.
3	王克华, 管凤宝, 夏祖虎, 等. 煤制氢低温甲醇洗装置改造方案研究[J]. 现代化工, 2023, 43(11): 231-235.
	WANG Kehua, GUAN Fengbao, XIA Zuhu, et al. Research on upgrading scheme for rectisol unit in a coal to hydrogen plant[J]. Modern Chemical Industry, 2023, 43(11): 231-235.
4	李希龙. 低温甲醇洗H₂S和CO₂吸收塔流程模拟与优化[J]. 石油石化绿色低碳, 2023, 8(1): 63-69.
	LI Xilong. Process simulation and optimization of rectisol H₂S and CO₂ absorption towers[J]. Green Petroleum & Petrochemicals, 2023, 8(1): 63-69.
5	张陆, 杨声. 低温甲醇洗碳捕集改进与优化[J]. 化工进展, 2022, 41(11): 6167-6175.
	ZHANG Lu, YANG Sheng. Improvement and optimization of carbon capture via Rectisol[J]. Chemical Industry and Engineering Progress, 2022, 41(11): 6167-6175.
6	FENG Yixiong, ZHAO Yuliang, ZHENG Hao, et al. Data-driven product design toward intelligent manufacturing: A review[J]. International Journal of Advanced Robotic Systems, 2020, 17(2): 172988142091125.
7	OLMEZ Elif, EGRIOGLU Erol, Eren BAS. Bootstrapped dendritic neuron model artificial neural network for forecasting[J]. Granular Computing, 2023, 8(6): 1689-1699.
8	LI Weiwei, SONG Yuncai. Artificial neural network model of catalytic coal gasification in fixed bed[J]. Journal of the Energy Institute, 2022, 105: 176-183.
9	丛迪. 乙烯生产过程能效分析与优化方法研究[D]. 北京: 北京化工大学, 2022.
	CONG Di. Research on energy efficiency analysis and optimization method for ethylene production process[D]. Beijing: Beijing University of Chemical Technology, 2022.
10	杨宇轩, 潘欣, 鄢烈祥, 等. 基于BP神经网络和列队竞争算法的低温甲醇洗过程参数优化[J]. 计算机与应用化学, 2013, 30(12): 1439-1443.
	YANG Yuxuan, PAN Xin, YAN Liexiang, et al. Parameters optimization in rectisol wash process based on BP neural network and line-up competition algorithm[J]. Computers and Applied Chemistry, 2013, 30(12): 1439-1443.
11	李英泽. 基于数据分析与流程模拟的低温甲醇洗操作优化[D]. 广州: 华南理工大学, 2021.
	LI Yingze. Operation optimization of the rectisol based on data analysis and process simulation [D]. Guangzhou: South China University of Technology, 2021.
12	李勇, 赵宇明. 基于贝叶斯优化算法和长短期记忆网络的PM_2.5浓度预测[J]. 流体测量与控制, 2023, 4(6): 14-17, 21.
	LI Yong, ZHAO Yuming. PM_2.5 concentration prediction based on Bayesian optimization algorithm and LSTM[J]. Fluid Measurement & Control, 2023, 4(6): 14-17, 21.
13	邱凯旋, 李佳. 基于贝叶斯优化和长短期记忆神经网络（BO-LSTM）的短期电力负荷预测[J]. 电力学报, 2022, 37(5): 367-373.
	QIU Kaixuan, LI Jia. Short-term power load forecasting based on Bayes optimization and long and short-term memory neural network(BO-LSTM)[J]. Journal of Electric Power, 2022, 37(5): 367-373.
14	WU Jia, LIU Xiyuan, CHEN Senpeng. Hyperparameter optimization through context-based meta-reinforcement learning with task-aware representation[J]. Knowledge-Based Systems, 2023, 260: 110160.
15	STIEGEL Gary J, MAXWELL Russell C. Gasification technologies: The path to clean, affordable energy in the 21st century[J]. Fuel Processing Technology, 2001, 71(1/2/3): 79-97.
16	李鹏飞. 连续重整装置的数据驱动建模和重点单元优化[D]. 大连: 大连理工大学, 2022.
	LI Pengfei. Data-driven modeling and key unit optimization of continuous reformer[D]. Dalian: Dalian University of Technology, 2022.
17	顾俊发, 许明阳, 马方圆, 等. 基于MIC的支持向量回归及其在化工过程中的应用[J]. 化工学报, 2021, 72(3): 1480-1486.
	GU Junfa, XU Mingyang, MA Fangyuan, et al. Support vector regression based on maximal information coefficient and its application in chemical industrial processes[J]. CIESC Journal, 2021, 72(3): 1480-1486.
18	高晓红, 李兴奇. 多元线性回归模型中无量纲化方法比较[J]. 统计与决策, 2022, 38(6): 5-9.
	GAO Xiaohong, LI Xingqi. Comparison of dimensionless methods in multiple linear regression models[J]. Statistics & Decision, 2022, 38(6): 5-9.
19	LIU Chuanlu, WANG Shuliang, YUAN Hanning, et al. Detecting unbiased associations in large data sets[J]. Big Data, 2022, 10(4): 337-355.
20	周家伟. 基于MIC-LSTM的水体连续缺失数据插补[J]. 长江信息通信, 2023, 36(3): 58-61.
	ZHOU Jiawei. Interpolation of water successive missing data based on MIC-LSTM[J]. Changjiang Information & Communications, 2023, 36(3): 58-61.
21	柳守诚, 王淳, 邹智辉, 等. 基于知识驱动及数据相关性的低压配电网户变关系识别[J]. 南昌大学学报（工科版）, 2023, 45(4): 392-398.
	LIU Shoucheng, WANG Chun, ZOU Zhihui, et al. Identification of household transformer relationship in low voltage distribution network based on knowledge driven and data correlation[J]. Journal of Nanchang University(Engineering & Technology), 2023, 45(4): 392-398.
22	潘艳秋, 李鹏飞, 高石磊, 等. 基于MIC筛选规则和BP神经网络的变换装置建模及产品预测[J]. 化工进展, 2022, 41(S1): 36-43.
	PAN Yanqiu, LI Pengfei, GAO Shilei, et al. Application of BP neural network based on MIC screening rules in modeling and product prediction of shift unit[J]. Chemical Industry and Engineering Progress, 2022, 41(S1): 36-43.
23	PAN Shiyuan, LIAO Qi, LIANG Yongtu. Multivariable sales prediction for filling stations via GA improved BiLSTM[J]. Petroleum Science, 2022, 19(5): 2483-2496.
24	张宜祥, 张玲华. 基于超参数优化的电力负荷预测模型研究[J]. 电子设计工程, 2024, 32(4): 37-42.
	ZHANG Yixiang, ZHANG Linghua. Research on power load forecasting model based on optimization of hyper-parameter[J]. Electronic Design Engineering, 2024, 32(4): 37-42.
25	LI Yucen Lily, RUDNER Tim G J, WILSON Andrew Gordon. A study of Bayesian neural network surrogates for Bayesian optimization[EB/OL]. 2023: 2305.20028. .
26	刘行, 王秋晨, 文韵豪, 等. 基于BO-LSTM的天然气处理厂负荷率预测模型[J]. 天然气与石油, 2023, 41(5): 122-130.
	LIU Xing, WANG Qiuchen, WEN Yunhao, et al. Forecasting model for load rate of natural gas treatment plant based on BO-LSTM model[J]. Natural Gas and Oil, 2023, 41(5): 122-130.
27	王樱达, 丁泽, 王延瓒, 等. 基于贝叶斯优化Bi-LSTM的刀具磨损状态监测模型[J]. 工具技术, 2023, 57(6): 133-137.
	WANG Yingda, DING Ze, WANG Yanzan, et al. Tool wear detection based on Bayesian optimized Bi-LSTM[J]. Tool Engineering, 2023, 57(6): 133-137.
28	刘智, 李欣雨, 李震, 等. 基于Bayes-LSTM的公路隧道围岩变形预测方法研究[J]. 中外公路, 2024, 44(1): 166-176.
	LIU Zhi, LI Xinyu, LI Zhen, et al. Prediction method of surrounding rock deformation of highway tunnels based on Bayes-LSTM[J]. Journal of China & Foreign Highway, 2024, 44(1): 166-176.
29	雷萌, 吕游, 魏玮, 等. 基于LSTM神经网络与贝叶斯优化的电站风机故障预警[J]. 热能动力工程, 2022, 37(8): 213-220.
	LEI Meng, You LYU, WEI Wei, et al. Fault warning of power plant fans based on long short-term memory neural network and Bayesian optimization[J]. Journal of Engineering for Thermal Energy and Power, 2022, 37(8): 213-220.

[1]	张迁, 刘鑫, 王冰, 徐晶, 曹晨熙. 复杂风速风向与事件树下储罐区多米诺事故分析[J]. 化工进展, 2025, 44(2): 1170-1182.
[2]	孙俭, 张海勇, 王成秀, 孙泽能, 蓝兴英, 高金森, 祝京旭. 基于k-means机器学习方法的气固循环流化床颗粒聚团特性[J]. 化工进展, 2025, 44(2): 625-634.
[3]	潘家保, 李贻良, 王锦. 磁流变脂热效应行为的分析与预测[J]. 化工进展, 2025, 44(1): 415-423.
[4]	戴征舒, 左元浩, 陈孝罗, 张犁, 赵根, 张学军, 张华. 机器学习在喷射器研究中的应用进展[J]. 化工进展, 2024, 43(S1): 1-12.
[5]	邹志云, 于蒙, 刘英莉. 基于LSTM和BP神经网络的间歇蒸馏过程工况预测[J]. 化工进展, 2024, 43(S1): 21-31.
[6]	陈王觅, 席北斗, 李鸣晓, 叶美瀛, 侯佳奇, 于承泽, 魏域芳, 孟繁华. 热解系统碳排放削减技术研究进展[J]. 化工进展, 2024, 43(S1): 479-503.
[7]	李依梦, 陈运全, 何畅, 张冰剑, 陈清林. 基于物理信息神经网络的甲烷无氧芳构化反应的正反问题[J]. 化工进展, 2024, 43(9): 4817-4823.
[8]	张佳鑫, 张淼, 戴一阳, 董立春. 面向实际化工过程故障诊断的强化深度卷积神经网络模型构建与应用[J]. 化工进展, 2024, 43(9): 4833-4844.
[9]	王亚男, 刘琳琳, 庄钰, 都健. 基于代理模型的环氧乙烷制乙二醇工艺优化同步热集成[J]. 化工进展, 2024, 43(9): 5234-5241.
[10]	丁路, 王培尧, 孔令学, 白进, 于广锁, 李文, 王辅臣. 煤气化过程反应模型研究进展[J]. 化工进展, 2024, 43(7): 3593-3612.
[11]	李凯, 魏鹤琳, 尹志凡, 左夏华, 于晓宇, 尹宏远, 杨卫民, 阎华, 安瑛. 基于GA-BP神经网络模型预测水基炭黑-胶原蛋白纳米流体热导率和黏度[J]. 化工进展, 2024, 43(7): 4138-4147.
[12]	郑锁祺, 詹凌霄, 陈恒, 李志浩, 王禹瑞, 赵宁, 吴昊, 杨林军. 基于混合模型的脱硫废水旁路蒸发系统能耗特性[J]. 化工进展, 2024, 43(6): 2968-2976.
[13]	张真, 张凡, 云祉婷. 绿氢在石化和化工行业的减碳经济性分析[J]. 化工进展, 2024, 43(6): 3021-3028.
[14]	吴琪, 白柏杨, 尹永杰, 马晓迅. 鄂尔多斯褐煤显微组分结构与其热解特性间的关系[J]. 化工进展, 2024, 43(5): 2370-2385.
[15]	李欣泽, 邹炜杰, 孙晨, 付璇, 陈潜, 袁亮, 王梓丞, 邢晓凯, 熊小琴, 郭良辉. 新疆油田某超临界CO₂管道安全停输时间预测[J]. 化工进展, 2024, 43(5): 2823-2833.

基于BO-LSTM的低温甲醇洗净化气CO₂含量预测

Prediction of CO₂ content in Rectisol purified gas based on BO-LSTM

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 12

参考文献 29

相关文章 15

编辑推荐

Metrics

本文评价