Integrated modelling method for tank-batch finished gasoline blending formulations

doi:10.16085/j.issn.1000-6613.2021-2262

Abstract

Abstract:

During the blending process of tank batches of finished gasoline,an integrated modeling method for finished gasoline blending general formula was proposed by considering the influence of the remaining oil at the bottom of the tank on product quality indicators, which based on an improved multi-kernel fuzzy C-means (MKFCM) and an extreme gradient boosting tree (XGBoost). Firstly, considering the differences of the blending components, an adaptive kernel parameter calculation method was presented to improve MKFCM, and then it was used for tank bottom oil cluster analysis in order to classify tank bottom oils with similar properties to the greatest extent batch-type. Secondly, XGBoost algorithm was selected on this basis. The sub-formulation model of each batch was established by taking the component proportion of residual oil at the bottom of each batch of tank and the expected quality index of the product as the input. When generating the formula, the membership vector of the current tank bottom residual oil was obtained based on the improved MKFCM and the above membership vectors were used to weight and fuse the sub formula model. Finally, the blended general formula of finished gasoline was obtained by multi-model integration. The experimental analysis by using the actual industrial data of an enterprise showed that compared with a single model or an integrated model without the improved MKFCM, the multi-model integrated formula based on the improved MKFCM-XGBoost possessed better prediction accuracy and generalization ability, and it was more suitable for tank type process of blending batches of finished gasoline.

Key words: blending formula, prediction, improve MKFCM, extreme gradient boosting (XGBoost), integration, remaining oil at the bottom of the tank, optimization

摘要：

考虑罐式批次成品汽油调和过程中罐底余油对产品质量指标的影响，本文提出了一种基于改进多核模糊C均值（multi-kernel fuzzy C-means，MKFCM）与极端梯度提升树（extreme gradient boosting，XGBoost）集成的成品汽油调和的通用配方建模方法。该方法首先考虑各调和组分的差异性，提出一种自适应核参数计算方法对MKFCM改进，并将其用于罐底油聚类分析，旨在最大程度划分出性质相近的罐底油批次类型；在此基础上择优选用XGBoost算法，以各批次罐底余油的组分比例和产品预期质量指标作为输入，建立各批次子配方模型；在配方生成时，基于改进MKFCM求得当前罐底余油的隶属度向量，并以此为权值对子配方模型进行加权融合，最终以多模型集成的方式得到了成品汽油调和的通用配方。经使用某企业实际工业数据进行实验分析，结果表明，较单一模型或未改进MKFCM的集成模型，基于改进MKFCM-XGBoost的多模型集成配方，预测精度和泛化能力均更优，更适合罐式批次成品汽油调和过程。

关键词: 调和配方, 预测, 改进多核模糊C均值, 极端梯度提升树, 集成, 罐底油, 优化

CLC Number:

TP273

LI Wei, RUAN Chenglong, WANG Xiaoming, LI Yajie, LIANG Chenglong. Integrated modelling method for tank-batch finished gasoline blending formulations[J]. Chemical Industry and Engineering Progress, 2022, 41(9): 4701-4712.

李炜, 阮成龙, 王晓明, 李亚洁, 梁成龙. 罐式批次成品汽油调和配方集成建模方法[J]. 化工进展, 2022, 41(9): 4701-4712.

Figures/Tables 12

References 39

1	宋倩倩, 慕彦君, 侯雨璇, 等. 中美两国石油化工产业实力对比分析[J]. 化工进展, 2020, 39(5): 1607-1619.
	SONG Qianqian, MU Yanjun, HOU Yuxuan, et al. Comparative analysis of the strength of petrochemical industry between China and USA[J]. Chemical Industry and Engineering Progress, 2020, 39(5): 1607-1619.
2	李炜, 王晓明, 蒋栋年, 等. 基于SHPSO-GA-BP的成品汽油调和中加氢汽油组分辛烷值的预测[J]. 化工学报, 2020, 71(7): 3191-3200.
	LI Wei, WANG Xiaoming, JIANG Dongnian, et al. Prediction of octane number of finished gasoline blend based on SHPSO-GA-BP[J]. CIESC Journal, 2020, 71(7): 3191-3200.
3	CASTRO P M. Source-based discrete and continuous-time formulations for the crude oil pooling problem[J]. Computers & Chemical Engineering, 2016, 93: 382-401.
4	TAIFOURIS M, MARTÍN M, MARTÍNEZ A, et al. Simultaneous optimization of the design of the product, process, and supply chain for formulated product[J]. Computers & Chemical Engineering, 2021, 152: 107384.
5	CASTRO P M. New MINLP formulation for the multiperiod pooling problem[J]. AIChE Journal, 2015, 61(11): 3728-3738.
6	LIU L Y. Molecular characterisation and modelling for refining processes[D]. University of Manchester, 2015
7	JAMRI M AL, LI J, SMITH R. Molecular characterisation of biomass pyrolysis oil and petroleum fraction blends[J]. Computers & Chemical Engineering, 2020, 140: 106906.
8	王通, 段泽文. 基于模糊评估自适应更新的油井动液面软测量建模[J]. 化工学报, 2019, 70(12): 4760-4769.
	WANG Tong, DUAN Zewen. Soft sensor modeling for dynamic liquid level of oil well based on fuzzy inference adaptive updating[J]. CIESC Journal, 2019, 70(12): 4760-4769.
9	何新礼, 谢莉, 杨慧中. 基于DP-RFR的多模型软测量建模[J]. 控制工程, 2020, 27(1): 64-69.
	HE Xinli, XIE Li, YANG Huizhong. Multi-model soft sensor modeling based on the DP-RFR method[J]. Control Engineering of China, 2020, 27(1): 64-69.
10	TARIQ T, KADDOUR M. A novel cluster head selection method based on HAC algorithm for energy efficient wireless sensor network[C]// IPAC ’15: Proceedings of the International Conference on Intelligent Information Processing, Security and Advanced Communication. 2015: 1-6.
11	刘聪, 谢莉, 杨慧中. 基于改进DPC的青霉素发酵过程多模型软测量建模[J]. 化工学报, 2021, 72(3): 1606-1615.
	LIU Cong, XIE Li, YANG Huizhong. Multi-model soft sensor development for penicillin fermentation process based on improved density peak clustering[J]. CIESC Journal, 2021, 72(3): 1606-1615.
12	吴小燕, 陈松灿. 联机核模糊C均值聚类方法[J]. 系统工程与电子技术, 2012, 34(12): 2599-2606.
	WU Xiaoyan, CHEN Songcan. Online kernel fuzzy C-means clustering algorithm[J]. Systems Engineering and Electronics, 2012, 34(12): 2599-2606.
13	张敏, 于剑. 基于划分的模糊聚类算法[J]. 软件学报, 2004, 15(6): 858-868.
	ZHANG Min, YU Jian. Fuzzy partitional clustering algorithms[J]. Journal of Software, 2004, 15(6): 858-868.
14	任喜伟, 何立风, 姚斌，等. 油水界测量过程中自适应阈值聚类优化算法[J]. 化工进展, 2019, 38(2): 779-789.
	REN Xiwei, HE Lifeng, YAO Bin, et al. Clustering optimization algorithm with adaptive threshold for oil-water interface detection process[J]. Chemical Industry and Engineering Progress, 2019, 38(2): 779-789.
15	KUMAR A, SHARMA K, SHARMA A. Genetically optimized fuzzy C-means data clustering of IoMT-based biomarkers for fast affective state recognition in intelligent edge analytics[J]. Applied Soft Computing, 2021, 109: 107525.
16	徐洋, 方洋旺, 伍友利, 等. 相似分布特性准则下的高斯混合项聚类-合并算法[J]. 国防科技大学学报, 2019, 41(4): 156-164.
	XU Yang, FANG Yangwang, WU Youli, et al. Cluster-merge method for the Gaussian mixture components based on the similarity distribution criterion[J]. Journal of National University of Defense Technology, 2019, 41(4): 156-164.
17	SILVA SOUZA M P DA, DE MENEZES E SILVA FILHO T, AMARAL G J A D, et al. Investigating different fitness criteria for swarm-based clustering[J]. International Journal of Business Intelligence and Data Mining, 2019, 15(1): 117.
18	田睿, 范祥祥, 戴影, 等. 核模糊C均值聚类算法优选BDS-3三频组合观测值[J]. 系统工程与电子技术, 2020, 42(3): 686-697.
	TIAN Rui, FAN Xiangxiang, DAI Ying, et al. Optimal triple-frequency combination observations for BDS-3 derived from a modified kernel-based fuzzy C-means clustering algorithm[J]. Systems Engineering and Electronics, 2020, 42(3): 686-697.
19	刘云, 刘富, 侯涛, 等. 优化核参数的模糊C均值聚类算法[J]. 吉林大学学报(工学版), 2016, 46(1): 246-251.
	LIU Yun, LIU Fu, HOU Tao, et al. Kernel-based fuzzy C-means clustering method based on parameter optimization[J]. Journal of Jilin University (Engineering and Technology Edition), 2016, 46(1): 246-251.
20	HUANG H C, CHUANG Y Y, CHEN C S. Multiple kernel fuzzy clustering[J]. IEEE Transactions on Fuzzy Systems, 2012, 20(1): 120-134.
21	ZENG S, WANG Z Y, HUANG R, et al. A study on multi-kernel intuitionistic fuzzy C-means clustering with multiple attributes[J]. Neurocomputing, 2019, 335: 59-71.
22	孙建平, 苑一方. 复杂过程的多模型建模方法研究[J]. 仪器仪表学报, 2011, 32(1): 132-137.
	SUN Jianping, YUAN Yifang. Multi-model modeling approach for complex process[J]. Chinese Journal of Scientific Instrument, 2011, 32(1): 132-137.
23	牛培峰, 刘超, 李国强, 等. 基于双层聚类与GSA-LSSVM的汽轮机热耗率多模型预测[J]. 电机与控制学报, 2016, 20(3): 90-95.
	NIU Peifeng, LIU Chao, LI Guoqiang, et al. Multi-model for turbine heat rate forecasting based on double layer clustering algorithm and GSA-LSSVM[J]. Electric Machines and Control, 2016, 20(3): 90-95.
24	张孙力, 杨慧中. 一种基于改进扩张搜索聚类算法的软测量建模方法[J]. 南京理工大学学报, 2017, 41(5): 574-580.
	ZHANG Sunli L, YANG Huizhong. Soft sensor modeling method based on improved expanding searching clustering algorithm[J]. Journal of Nanjing University of Science and Technology, 2017, 41(5): 574-580.
25	葛祥振. 基于仿射传播聚类和高斯过程回归的软测量建模研究[D]. 无锡: 江南大学, 2017.
	GE Xiangzhen. Research on soft sensor modeling based on affinity propagation clustering and Gaussian process regression[D]. Wuxi: Jiangnan University, 2017.
26	许言路, 张建森, 吉星, 等. 基于多模型融合神经网络的短期负荷预测[J]. 控制工程, 2019, 26(4): 619-624.
	XU Yanlu, ZHANG Jiansen, JI Xing, et al. Research on short-term load forecasting method based on multi-model fusion neural network[J]. Control Engineering of China, 2019, 26(4): 619-624.
27	ACCARINO G, CHIARELLI M, FIORE S, et al. A multi-model architecture based on long short-term memory neural networks for multi-step sea level forecasting[J]. Future Generation Computer Systems, 2021, 124: 1-9.
28	CHEN T Q, GUESTRIN C. XGBoost: a scalable tree boosting system[C]// KDD ’16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2016: 785-794.
29	DING Y, FAN L X, LIU X. Analysis of feature matrix in machine learning algorithms to predict energy consumption of public buildings[J]. Energy and Buildings, 2021, 249: 111208.
30	朱礼涛, 欧阳博, 张希宝，等. 机器学习在多相反应中的应用进展[J]. 化工进展, 2021, 40(4): 1699-1714.
	ZHU Litao, OUYANG Bo, ZHANG Xibao, et al. Progress on application of machine learning to multiphase reactors[J]. Chemical Industry and Engineering Progress, 2021, 40(4): 1699-1714.
31	QIU Y, GARG D, KIM S M, et al. Machine learning algorithms to predict flow boiling pressure drop in mini/micro-channels based on universal consolidated data[J]. International Journal of Heat and Mass Transfer, 2021, 178: 121607.
32	陈延展, 胡浩, 任紫畅, 等. 基于XGBoost和改进灰狼优化算法的催化裂化汽油精制装置的辛烷值损失模型分析[J/]. 石油学报(石油加工), 2022, 38(1): 208-219.
	CHEN Yanzhan, HU Hao, REN Zichang, et al. Model analysis of gasoline octane loss in catalytic cracking post-refining unit based on XGBoost and improved gray wolf optimization algorithm[J]. Acta Petrolei Sinica (Petroleum Processing Section), 2022, 38(1): 208-219.
33	何仁初, 陈海泉, 杨超文. 面向国Ⅵ标准的汽油调合优化技术探讨与分析[J]. 化工进展, 2018, 37(3): 962-969.
	HE Renchu, CHEN Haiquan, YANG Chaowen. Discussion and analysis on optimization technology of gasoline blending for Chinese national standard Ⅵ[J]. Chemical Industry and Engineering Progress, 2018, 37(3): 962-969.
34	RUSIN M H, CHUNG H S, MARSHALL J F. A “transformation” method for calculating the research and motor octane numbers of gasoline blends[J]. Industrial & Engineering Chemistry Fundamentals, 1981, 20(3): 195-204.
35	李春艳. 基于信息熵与多核学习改进的多视角FCM聚类的研究[D]. 苏州: 苏州大学, 2020.
	LI Chunyan. Research on improved multi-view fuzzy C-means clustering based on information entropy and multiple kernel learning[D]. Suzhou: Soochow University, 2020.
36	周绍磊, 廖剑, 史贤俊. 基于Fisher准则和最大熵原理的SVM核参数选择方法[J]. 控制与决策, 2014, 29(11): 1991-1996.
	ZHOU Shaolei, LIAO Jian, SHI Xianjun. SVM parameters selection method based on Fisher criterion and maximum entropy principle[J]. Control and Decision, 2014, 29(11): 1991-1996.
37	ZHANG Yaling, HAN Jin. Differential privacy fuzzy C-means clustering algorithm based on Gaussian kernel function[J]. PLoS one, 2021, 16(3): e0248737.
38	ZHOU S H, ZHU E, LIU X W, et al. Subspace segmentation-based robust multiple kernel clustering[J]. Information Fusion, 2020, 53: 145-154.
39	TIBSHIRANI R, WALTHER G, HASTIE T. Estimating the number of clusters in a data set via the gap statistic[J]. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 2001, 63(2): 411-423.

算法	主要参数	参数设置
改进 MKFCM	整体核参数σ	0.005
	聚类个数C	3
	平滑因子m	2
	迭代结束条件	‖ U （i）- U （i-1）‖<0.0001

算法	主要参数	参数设置
改进 MKFCM	整体核参数σ	0.005
	聚类个数C	3
	平滑因子m	2
	迭代结束条件	‖ U （i）- U （i-1）‖<0.0001

模型	主要参数	MTBE	乙苯	甲苯	搜索范围
XGBoost1/	n_estimators	2182/2855/7902/2300	200/1200/15002300	350/1300/2960/2300	200~10000
XGBoost1/	colsample_bytree	0.77/0.71/0.71/1	1/1/0.77/1	0.25/0.3/0.9/1	0~1
XGBoost2/	colsample_bylevel	1	1	1/1/0.9/1	0~1
XGBoost2/	colsample_bynode	1	1	1/1/0.5/1	0~1
XGBoost3/	η	0.14/0.05/0.05/0.3	0.16/0.05/0.14/0.3	0.3/0.3/0.7/0.3	0~1
XGBoost	max_depth	7/8/10/6	15/13/16/6	8/8/6/6	6~20

模型	主要参数	MTBE	乙苯	甲苯	搜索范围
XGBoost1/	n_estimators	2182/2855/7902/2300	200/1200/15002300	350/1300/2960/2300	200~10000
XGBoost1/	colsample_bytree	0.77/0.71/0.71/1	1/1/0.77/1	0.25/0.3/0.9/1	0~1
XGBoost2/	colsample_bylevel	1	1	1/1/0.9/1	0~1
XGBoost2/	colsample_bynode	1	1	1/1/0.5/1	0~1
XGBoost3/	η	0.14/0.05/0.05/0.3	0.16/0.05/0.14/0.3	0.3/0.3/0.7/0.3	0~1
XGBoost	max_depth	7/8/10/6	15/13/16/6	8/8/6/6	6~20

样本编号	MKFCM算法隶属度			改进MKFCM算法隶属度			所属类别
样本编号	Ⅰ类	Ⅱ类	Ⅲ类	Ⅰ类	Ⅱ类	Ⅲ类	所属类别
6	0.33301	0.33329	0.33368	0.36035	0.54604	0.09359	3/2
55	0.33309	0.33369	0.33320	0.30645	0.34325	0.35028	2/3
116	0.33367	0.33315	0.33316	0.38301	0.47266	0.14431	1/2
147	0.33332	0.33338	0.33328	0.35798	0.34124	0.30077	2/1