Research on multi-parameter optimization method based on parallel EGO and surrogate-assisted model

doi:10.11949/0438-1157.20221345

Abstract

Abstract:

Process simulation and optimization have the characteristics of high-dimensional and non-linear, which makes the simulation calculation difficult to converge. Excessive solution time is one of the bottlenecks during scheduling and operation optimization. Using surrogate models to replace mechanistic models is an effective way to reduce computational complexity and ensure the accuracy of results. Though the Kriging surrogate model has a stronger nonlinear approximation, it is still difficult to deal with high-dimensional problems. Therefore, this paper investigates the parallel EGO (efficient global optimization) algorithm integrated with the surrogate model and applies the model to typical chemical process. The parallel EGO algorithm is based on the prediction function and error function of the Kriging surrogate model. First, an analytical expression is derived from data sample combined by the probability density function and cumulative distribution function. Then the surrogate model is updated by new sample points obtained by PEI (pseudo expected improvement) criterion. Finally, combined with the improved differential evolution algorithm, the optimization parameters are searched globally. Under the premise of ensuring the accuracy of the results, the algorithm in this paper is compared with other optimization algorithms. Taking eight multi-peaked test functions as a test case, it is found that the convergence speed of the algorithm is improved by 85%. Then, it is applied to the study case of the two-stage ammonia absorption refrigeration process. The results show that the simulation error is less than 0.01%, and the optimization time is reduced from 9846 s to 3705 s.

Key words: surrogate model, parallel efficient global optimization algorithm, multi-parameter optimization, process simulation

摘要：

化工流程模拟优化问题常常具有高维、非线性的特点，使得仿真计算难以收敛。过长的求解时间是调度优化和运行优化的主要瓶颈之一。采用代理模型对机理模型进行替代是降低计算复杂度、保证结果准确性的有效途径。Kriging代理模型具有较强的非线性近似性，但处理高维问题依然较为困难。因此，本文研究并行EGO(efficient global optimization)算法与代理模型集成，并将模型应用于化工过程。并行EGO算法以Kriging代理模型的预测函数和误差函数为基础，先推导出样本分布概率密度函数与累积分布函数相结合的解析表达式；然后通过PEI(pseudo expected improvement)准则得到新的样本点以更新代理模型；最后结合改进的差分进化算法对优化参数进行全局搜索。在保证结果准确性的前提下，将本文算法与其他优化算法进行比较。8个多峰测试函数的测试结果表明，该算法的收敛速度提高了85%。然后将其应用于双级氨吸收制冷过程的模拟，结果表明该方法的模拟误差小于0.01%，优化时间从9846 s缩短至3705 s。

关键词: 代理模型, 并行EGO算法, 多参数优化, 流程模拟

CLC Number:

TQ 062.2

Xuerong GU, Shuoshi LIU, Siyu YANG. Research on multi-parameter optimization method based on parallel EGO and surrogate-assisted model[J]. CIESC Journal, 2023, 74(3): 1205-1215.

顾学荣, 刘硕士, 杨思宇. 基于并行EGO和代理模型辅助的多参数优化方法研究[J]. 化工学报, 2023, 74(3): 1205-1215.

Figures/Tables 11

Table 1 Parameters related to Hartman3 and Hartman6

函数	参数
Hartman3	$α i j = 3.0 10.0 30.0 0.1 10.0 35.0 3.0 10.0 30.0 0.1 10.0 35.0$
Hartman3	$p i j = 0.3689 0.1170 0.2673 0.4699 0.4378 0.7470 0.1091 0.8732 0.5547 0.03815 0.5743 0.8828$
Hartman6	$α i j = 10.00 3.00 17.00 3.50 1.70 8.00 0.05 10.00 17.00 0.10 8.00 14.00 3.00 3.50 1.70 10.00 17.00 8.00 17.00 8.00 0.05 10.00 0.10 14.00$
Hartman6	$p i j = 0.1312 0.1696 0.5569 0.0124 0.8283 0.5886 0.2329 0.4135 0.8307 0.3736 0.1004 0.9991 0.2348 0.1451 0.3522 0.2883 0.3047 0.6650 0.4047 0.8828 0.8732 0.5743 0.1091 0.0381$

Table 1 Parameters related to Hartman3 and Hartman6

函数	参数
Hartman3	$α i j = 3.0 10.0 30.0 0.1 10.0 35.0 3.0 10.0 30.0 0.1 10.0 35.0$
Hartman3	$p i j = 0.3689 0.1170 0.2673 0.4699 0.4378 0.7470 0.1091 0.8732 0.5547 0.03815 0.5743 0.8828$
Hartman6	$α i j = 10.00 3.00 17.00 3.50 1.70 8.00 0.05 10.00 17.00 0.10 8.00 14.00 3.00 3.50 1.70 10.00 17.00 8.00 17.00 8.00 0.05 10.00 0.10 14.00$
Hartman6	$p i j = 0.1312 0.1696 0.5569 0.0124 0.8283 0.5886 0.2329 0.4135 0.8307 0.3736 0.1004 0.9991 0.2348 0.1451 0.3522 0.2883 0.3047 0.6650 0.4047 0.8828 0.8732 0.5743 0.1091 0.0381$

Table 2 Test functions

测试函数	变量维度	初始样本点数	最大迭代次数
Six-hump	2	20	400
Branin	2	20	400
GoldPrice	2	20	400
Hartman3	3	30	400
Shekekl5	4	40	400
Shekekl7	4	40	400
Shekekl10	4	40	400
Hartman6	6	60	400

Fig.1 Iteration history curves of different optimization algorithms on test functions

Table 3 The number of iterations required to converge to the optimal solution

测试函数	变量维度	EGO算法	并行EGO算法
测试函数	变量维度	q=1	q=2	q=4	q=6	q=8	q=10
Six-hump	2	29.4	16.2	9.0	6.9	6.1	4.4
Branin	2	31.9	14.5	10.8	7.5	6.2	5.3
GoldPrice	2	97.8	47.3	24.3	21.2	15.6	11.6
Hartman3	3	16.0	14.0	9.0	7.0	6.0	4.0
Shekekl5	4	67.9	37.9	17.7	13.7	10.4	9.6
Shekekl7	4	165.0	54.1	33.5	19.8	17.2	13.0
Shekekl10	4	84.6	43.3	26.5	17.0	12.6	10.9
Hartman6	6	28.7	12.6	8.3	6.9	5.7	4.9

Fig.2 Running time of different multi-peak test functions

Fig.3 Schematic diagram of two-stage ammonia absorption refrigeration process

Table 4 Optimization variables and their constraints on ammonia absorption refrigeration

符号	物理意义	下限	上限
x₁	氨气( $N H 3$ )进料流量/（kmol/h）	5000	7000
x₂	闪蒸器气相分数	0.05	0.35
x₃	闪蒸器操作压力/kPa	1.0	1.4
x₄	换热器1换热温度/℃	40	50
x₅	泵压(泵1)/kPa	1450	1550
x₆	预热器换热温度/℃	10200	10700
x₇	精馏塔回流比	0.4	0.6
x₈	精馏塔塔顶采出率	0.45	0.65
x₉	过冷器换热温度/℃	10	15
x₁₀	压力调节阀/kPa	70	80
x₁₁	压缩机排气压力/kPa	100	120
x₁₂	换热器2换热温度/℃	35	45
x₁₃	吸收器换热温度/℃	40	50
x₁₄	泵压(泵2)/kPa	700	900
x₁₅	出口氨溶液(S)分流率	0.01	0.05

Table 4 Optimization variables and their constraints on ammonia absorption refrigeration

符号	物理意义	下限	上限
x₁	氨气( $N H 3$ )进料流量/（kmol/h）	5000	7000
x₂	闪蒸器气相分数	0.05	0.35
x₃	闪蒸器操作压力/kPa	1.0	1.4
x₄	换热器1换热温度/℃	40	50
x₅	泵压(泵1)/kPa	1450	1550
x₆	预热器换热温度/℃	10200	10700
x₇	精馏塔回流比	0.4	0.6
x₈	精馏塔塔顶采出率	0.45	0.65
x₉	过冷器换热温度/℃	10	15
x₁₀	压力调节阀/kPa	70	80
x₁₁	压缩机排气压力/kPa	100	120
x₁₂	换热器2换热温度/℃	35	45
x₁₃	吸收器换热温度/℃	40	50
x₁₄	泵压(泵2)/kPa	700	900
x₁₅	出口氨溶液(S)分流率	0.01	0.05

Fig.4 Comparison between predicted and actual values of Kriging model

Fig.5 Optimization iteration curves for different optimization algorithms

Table 5 The results of parameter optimization

变量	q=1	q=2	q=4	q=6	q=8	q=10
x₁	7000.00	5954.92	5959.70	5950.70	5952.17	5953.89
x₂	0.16	0.20	0.24	0.14	0.13	0.17
x₃	800.67	807.86	806.46	804.51	804.41	803.06
x₄	42.79	46.63	48.87	46.58	47.45	48.16
x₅	1550.00	1497.80	1507.33	1506.61	1500.00	1497.38
x₆	10.00	5.92	7.63	6.34	5.95	5.71
x₇	0.47	0.48	0.45	0.51	0.46	0.52
x₈	0.65	0.53	0.61	0.57	0.57	0.56
x₉	10.00	11.95	12.61	10.00	10.00	10.92
x₁₀	70.00	74.51	70.67	76.00	71.21	71.00
x₁₁	100.24	110.65	100.00	107.04	106.37	108.15
x₁₂	43.83	39.79	42.72	41.98	45.00	41.01
x₁₃	50.00	42.41	40.00	45.80	44.00	40.42
x₁₄	800.00	819.71	813.15	816.92	806.09	808.41
x₁₅	0.04	0.04	0.05	0.04	0.03	0.04
绝对误差	0.1566	0.0041	0.0010	0.00058	0.00044	0.00016

Fig.6 Running time of parallel EGO optimization algorithms

References 33

1	杨路, 刘硕士, 罗小艳, 等. MTO烯烃分离过程的多目标操作优化[J]. 化工学报, 2020, 71(10): 4720-4732.
	Yang L, Liu S S, Luo X Y, et al. Multi-objective operation optimization of olefin separation process for MTO plant[J]. CIESC Journal, 2020, 71(10): 4720-4732.
2	陈志杰, 金晓明. 精馏塔系的软测量建模与优化策略研究[J]. 计算机与应用化学, 2015, 32(3): 261-265.
	Chen Z J, Jin X M. Research of soft sensor and optimization on distillation column system[J]. Computers and Applied Chemistry, 2015, 32(3): 261-265.
3	Javaloyes-Antón J, Caballero J A, Ruiz-Femenia R. Rigorous design of complex distillation columns using process simulators and the particle swarm optimization algorithm[J]. Industrial & Engineering Chemistry Research, 2013, 52(44): 15621-15634.
4	Jiang P, Du W L. Multi-objective modeling and optimization for scheduling of cracking furnace systems[J]. Chinese Journal of Chemical Engineering, 2017, 25(8): 992-999.
5	Inamdar S V, Gupta S K, Saraf D N. Multi-objective optimization of an industrial crude distillation unit using the Elitist non-dominated sorting genetic algorithm[J]. Chemical Engineering Research and Design, 2004, 82(5): 611-623.
6	刘涛, 苏成利. 气体分馏装置的流程模拟及多目标优化[J]. 石油化工高等学校学报, 2013, 26(1): 76-80.
	Liu T, Su C L. Simulation analysis and multi-objective optimization of gas fractionation unit[J]. Journal of Petrochemical Universities, 2013, 26(1): 76-80.
7	Reddy P S, Rani K Y, Patwardhan S C. Multi-objective optimization of a reactive batch distillation process using reduced order model[J]. Computers & Chemical Engineering, 2017, 106: 40-56.
8	Tang M, Liu Y M, Durlofsky L J. A deep-learning-based surrogate model for data assimilation in dynamic subsurface flow problems[J]. Journal of Computational Physics, 2020, 413: 109456.
9	Slot R M M, Sørensen J D, Sudret B, et al. Surrogate model uncertainty in wind turbine reliability assessment[J]. Renewable Energy, 2020, 151: 1150-1162.
10	范才彬. 基于稀疏表达的多项式响应面模型及其仿真应用研究[D]. 武汉: 华中科技大学, 2015.
	Fan C B. Sparsity-promoting polynomial response surface and its application in simulation[D]. Wuhan: Huazhong University of Science and Technology, 2015.
11	龙腾, 刘建, WANG G Gary, 等. 基于计算试验设计与代理模型的飞行器近似优化策略探讨[J]. 机械工程学报, 2016, 52(14): 79-105.
	Long T, Liu J, Wang G G, et al. Discuss on approximate optimization strategies using design of computer experiments and metamodels for flight vehicle design[J]. Journal of Mechanical Engineering, 2016, 52(14): 79-105.
12	Eason J, Cremaschi S. Adaptive sequential sampling for surrogate model generation with artificial neural networks[J]. Computers & Chemical Engineering, 2014, 68: 220-232.
13	张琳, 汪廷华, 周慧颖. 基于群智能算法的SVR参数优化研究进展[J]. 计算机工程与应用, 2021, 57(16): 50-64.
	Zhang L, Wang T H, Zhou H Y. Research progress on parameter optimization of SVR based on swarm intelligence algorithm[J]. Computer Engineering and Applications, 2021, 57(16): 50-64.
14	Zhan D W, Qian J C, Cheng Y S. Pseudo expected improvement criterion for parallel EGO algorithm[J]. Journal of Global Optimization, 2017, 68(3): 641-662.
15	González-Gorbeña E, Qassim R Y, Rosman P. Optimisation of hydrokinetic turbine array layouts via surrogate modelling[J]. Renewable Energy, 2016, 93: 45-57.
16	Yu H, Tan Y, Sun C, et al. An adaptive model selection strategy for surrogate-assisted particle swarm optimization algorithm[C]//2016 IEEE Symposium Series on Computational Intelligence (SSCI). Los Alamitos: IEEE Computer Society Press, 2016: 1-8.
17	Liau L C K, Yang T C K, Tsai M T. Expert system of a crude oil distillation unit for process optimization using neural networks[J]. Expert Systems with Applications, 2004, 26(2): 247-255.
18	Singh P, Couckuyt I, Elsayed K, et al. Shape optimization of a cyclone separator using multi-objective surrogate-based optimization[J]. Applied Mathematical Modelling, 2016, 40(5/6): 4248-4259.
19	祁荣宾, 何镓梁, 靳文浩, 等. 基于Kriging代理模型的PX氧化过程优化[J]. 华东理工大学学报(自然科学版), 2012, 38(6): 712-717.
	Qi R B, He J L, Jin W H, et al. PX oxidation process optimization based on Kriging surrogate model[J]. Journal of East China University of Science and Technology (Natural Science Edition), 2012, 38(6): 712-717.
20	刘思乐, 王修纲, 吴静, 等. 基于Kriging代理模型的复杂精馏系统操作参数优化[J]. 石油炼制与化工, 2020, 51(1): 114-120.
	Liu S L, Wang X G, Wu J, et al. Operation parameter optimization of complex distillation system based on Kriging surrogate[J]. Petroleum Processing and Petrochemicals, 2020, 51(1): 114-120.
21	Ge X L, Yuan X G, Ao C, et al. Simulation based approach to optimal design of dividing wall column using random search method[J]. Computers & Chemical Engineering, 2014, 68: 38-46.
22	Jones D R, Schonlau M, Welch W J. Efficient global optimization of expensive black-box functions[J]. Journal of Global Optimization, 1998, 13(4): 455-492.
23	Krige D. A statistical approach to some basic mine valuation problems on the Witwatersrand[J]. Journal of the Southern African Institute of Mining and Metallurgy, 1951, 52(6): 119-139.
24	Sacks J, Welch W, Mitchell T J, et al. Design and analysis of computer experiments: rejoinder[J]. Statistical Science, 1989, 4: 433-435.
25	Cressie N. The origins of Kriging[J]. Mathematical Geology, 1990, 22(3): 239-252.
26	ForresterA I J, SóbesterA, KeaneA J. Engineering Design via Surrogate Modelling: A Practical Guide[M]. Chichester, West Sussex, England: J. Wiley, 2008.
27	Xing J, Luo Y, Gao Z. A global optimization strategy based on the Kriging surrogate model and parallel computing[J]. Structural and Multidisciplinary Optimization, 2020, 62(1): 405-417.
28	Toal D J J, Bressloff N W, Keane A J. Kriging hyperparameter tuning strategies[J]. AIAA Journal, 2008, 46(5): 1240-1252.
29	Kalagnanam J R, Diwekar U M. An efficient sampling technique for off-line quality control[J]. Technometrics, 1997, 39(3): 308-319.
30	Lophaven S N, Nielsen H B, Søndergaard J. DACE: a Matlab Kriging Toolbox[M]. Lyngby, Denmark: IMM, Informatics and Mathematical Modelling, The Technical University of Denmark, 2002.
31	Viana F A C, Haftka R T, Watson L T. Efficient global optimization algorithm assisted by multiple surrogate techniques[J]. Journal of Global Optimization, 2013, 56(2): 669-689.
32	Storn R, Price K. Differential evolution—a simple and efficient heuristic for global optimization over continuous spaces[J]. Journal of Global Optimization, 1997, 11(4): 341-359.
33	Liu S, Qian Y, Li D, et al. Multi-scenario scheduling optimization for a novel double-stage ammonia absorption refrigeration system incorporating an organic Rankine cycle[J]. Energy Conversion and Management, 2022, 270: 116170.

[1]	Ruihang ZHANG, Pan CAO, Feng YANG, Kun LI, Peng XIAO, Chun DENG, Bei LIU, Changyu SUN, Guangjin CHEN. Analysis of key parameters affecting product purity of natural gas ethane recovery process via ZIF-8 nanofluid [J]. CIESC Journal, 2023, 74(8): 3386-3393.
[2]	Xin ZHANG, Li ZHOU, Shihui WANG, Xu JI, Kexin BI. Integrated optimization of refinery hydrogen networks with crude oil properties fluctuations [J]. CIESC Journal, 2022, 73(4): 1631-1646.
[3]	Senshan CAO, Feng XU, Xionglin LUO. Process simulation of stream circulation system based on stability: [J]. CIESC Journal, 2022, 73(3): 1256-1269.
[4]	Qian LIU, Xianglan ZHANG, Zhiping LI, Zhuoqi LI, Hong YU. Multiscale screening of ionic liquids as extractive solvents for oil-hydroxybenzene separation [J]. CIESC Journal, 2022, 73(11): 5011-5024.
[5]	Li LIU, Peng JIANG, Wei WANG, Tonghuan ZHANG, Liwen MU, Xiaohua LU, Jiahua ZHU. Coupling process simulation and random forest model for analyzing and predicting biomass-to-hydrogen conversion [J]. CIESC Journal, 2022, 73(11): 5230-5239.
[6]	Kefan ZHAO, Shengkun JIA, Yiqing LUO, Xigang YUAN. Optimal design for dividing wall column using online Kriging surrogate model-based optimization method [J]. CIESC Journal, 2022, 73(1): 332-341.
[7]	Zixuan HUANG, Huan CHEN, Hai LI, Minglong WANG, Guangjin CHEN, Bei LIU. Process simulation and energy consumption analysis of CO₂/N₂ pilot-scale separation using ZIF-8 slurry [J]. CIESC Journal, 2022, 73(1): 322-331.
[8]	Yuanxin FANG, Wu XIAO, Xiaobin JIANG, Xiangcun LI, Gaohong HE, Xuemei WU. Process design and simulation of membrane separation coupled with CO₂ electrocatalytic hydrogenation to formic acid [J]. CIESC Journal, 2021, 72(9): 4740-4749.
[9]	Jianyuan XU, Yanyang WU, Jumei XU, Yangfeng PENG. Study on vapor-liquid equilibria and distillation simulation of 1,3,5-trimethylbenzene-1,2,4-trimethylbenzene and 1,3,5-trimethylbenzene-2-ethyltoluene at 2 kPa [J]. CIESC Journal, 2021, 72(9): 4504-4510.
[10]	Zhenxing SONG, Xianbao CUI, Ying ZHANG, Xuemei ZHANG, Jie HE, Tianyang FENG, Jixiao WANG. Synthesis of n-hexyl acetate via reactive distillation catalyzed by mixed ionic liquids [J]. CIESC Journal, 2021, 72(8): 4155-4165.
[11]	CAO Jian, FENG Xin, JI Xiaoyan, LU Xiaohua. Study on the theoretical limit performance of multi-pressure evaporation ORC based on zeotropic mixture [J]. CIESC Journal, 2021, 72(7): 3780-3787.
[12]	LU Zhibin, XIE Xing, LU Sida, HE Chang, ZHANG Bingjian, CHEN Qinglin. Surrogate model-based optimal design of multi-stage nanofiltration separation system for saline wastewater [J]. CIESC Journal, 2021, 72(3): 1400-1408.
[13]	LU Zhibin, QU Jinghui, LIU Hua, HE Chang, ZHANG Bingjian, CHEN Qinglin. Surrogate modeling for physical fields of heat transfer processes based on physics-informed neural network [J]. CIESC Journal, 2021, 72(3): 1496-1503.
[14]	Hui YANG, Wenhao DAI, Rongxiu LU, Jianyong ZHU. Simulation of rare earth extraction process based on separation coefficient correction [J]. CIESC Journal, 2020, 71(7): 3180-3190.
[15]	Yonghou XIAO, Hongyan XIAO, Benyuan LI, Jianliang QIN, Shuang QIU, Gaohong HE. Optimization of helium/methane adsorption separation process based on Aspen Adsorption simulation [J]. CIESC Journal, 2019, 70(7): 2556-2563.