组合加点准则的代理辅助多目标粒子群优化

doi:10.16180/j.cnki.issn1007-7820.2022.12.004

Abstract

Abstract:

In view of the problem of low initial optimization efficiency and poor model accuracy when constructing surrogate models with small sample data, a surrogate-assisted multi-objective particle swarm optimization algorithm based on the combined infill sampling criterion is proposed in this study. The algorithm combines the Kriging model and the radial basis function network model into a heterogeneous ensemble model through the weighted average method, and uses the combined infill sampling criterion of the improved expectation criterion and the minimum surrogate model prediction criterion to manage the surrogate model to speed up the convergence of the model. In addition, the proposed algorithm adopts the actual objective function to evaluate the sample points added in each iteration, and updates the surrogate model to increase the model accuracy. The experimental results show that compared with the non-surrogate model algorithm, the proposed algorithm reduces the evaluation times of the fitness function by 10 times, which proves that the proposed algorithm can improve the optimization efficiency and accuracy of the surrogate model, and achieve a better balance between exploration and development.

Key words: surrogate model, Kriging model, radial basis function network model, heterogeneous ensemble, combined infill sampling criterion, fitness function evaluation, model management, multi-objective particle swarm optimization algorithm

CLC Number:

TP18

CHEN Wanfen,WANG Yujia,LIN Weixing. Surrogate Assisted Multi-Objective Particle Swarm Optimization Based on Combined Infill Sampling Criterion[J].Electronic Science and Technology, 2022, 35(12): 26-34.

Figures/Tables 11

Figure 1.

Figure 2.

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Figure 3.

Figure 4.

References 33

[1]	Jin Y, Wang H, Chugh T, et al. Data-driven evolutionary optimization:An overview and case studies[J]. IEEE Transactions on Evolutionary Computation, 2018, 23(3):442-458. doi: 10.1109/TEVC.2018.2869001
[2]	Lian Y, Liou M S. Multi-objective optimization using coupled response surface model and evolutionary algorithm[J]. Aeronautics and Space Flight, 2005, 43(6):1316-1325.
[3]	蓝机满. 基于径向基神经网络的粒子群表面缺陷识别算法[J]. 电子科技, 2019, 32(5):92-95.
	Lan Jiman. Particle swarm optimization surface defect recognition algorithm based on radial basis neural network[J]. Electronic Science and Technology, 2019, 32(5):92-95.
[4]	王丽娟, 丁世飞. 一种粒子群优化的SVM-ELM模型[J]. 计算机科学与探索, 2019, 13(4):657- 665. doi: 10.3778/j.issn.1673-9418.1806015
	Wang Lijuan, Ding Shifei. SVM-ELM model based on particle swarm optimization[J]. Journal of Frontiers of Computer Science and Technology, 2019, 13(4):657-665. doi: 10.3778/j.issn.1673-9418.1806015
[5]	Ginsbourger D, Le Riche R, Carraro L. Kriging is well-suited to parallelize optimization[M]. Heidelberg:Springer, 2010.
[6]	Chugh T, Jin Y, Miettinen K, et al. A surrogate assisted reference vector guided evolutionary algorithm for computationally expensive many objective optimization[J]. IEEE Transactions on Evolutionary Computation, 2018, 22(1):129-142. doi: 10.1109/TEVC.2016.2622301
[7]	朱苗苗, 潘伟杰, 刘翔, 等. 基于BP神经网络代理模型的交互式遗传算法[J]. 计算机工程与应用, 2020, 56(2):146-151. doi: 10.3778/j.issn.1002-8331.1810-0089
	Zhu Miaomiao, Pan Weijie, Liu Xiang, et al. Interactive genetic algorithm based on BP neural network and user cognitive surrogate model[J]. Computer Engineering and Applications, 2020, 56(2):146-151. doi: 10.3778/j.issn.1002-8331.1810-0089
[8]	Forrester A I J, Keane A J. Recent advances in surrogate-based optimization[J]. Progress in Aerospace Sciences, 2009, 45(1):50-79. doi: 10.1016/j.paerosci.2008.11.001
[9]	Wang X L, Jin Y C, Sebastian S, et al. An adaptive bayesian approach to surrogate assisted evolutionary multi-objective optimization[J]. Information Sciences, 2020, 51(9):317-331.
[10]	Ruan X R, Li K, Derbel B, et al. Surrogate assisted evolutionary algorithm for medium scale multi-objective optimization problems[C]. New York: Proceedings of the Genetic and Evolutionary Computation Conference, 2020.
[11]	Zhou Z, Ong Y S, Nair P B, et al. Combining global and local surrogate models to accelerate evolutionary optimization[J]. IEEE Transactions on Systems,Man,and Cybernetics Part C, 2006, 37(1):66-76. doi: 10.1109/TSMCC.2005.855506
[12]	Han Z H, Zhang K S. Surrogate-based optimization[J]. Real-World Applications of Genetic Algorithms, 2012, 34(3):1-12.
[13]	Guo D, Jin Y, Ding J, et al. Heterogeneousensemble-based infill criterion for evolutionary multi-objective optimization of expensive problems[J]. IEEE Transactions on Cybernetics, 2018, 49(3):1012-1025. doi: 10.1109/TCYB.2018.2794503
[14]	Knowles J. ParEGO:A hybrid algorithm with on-line landscape approximation for expensive multi-objective optimization problems[J]. IEEE Transactions on Evolutionary Computation, 2006, 10(1):50-66. doi: 10.1109/TEVC.2005.851274
[15]	Zhang Q, Liu W, Tsang E, et al. Expensive multi-objective optimization by MOEA/D with gaussian process model[J]. IEEE Transactions on Evolutionary Computation, 2010, 14(3):456-474. doi: 10.1109/TEVC.2009.2033671
[16]	Liu B, Zhang Q, Gielen G G E. A gaussian process surrogate model assisted evolutionary algorithm for medium scale expensive optimization problems[J]. IEEE Transactions on Evolutionary Computation, 2013, 18(2):180-192. doi: 10.1109/TEVC.2013.2248012
[17]	Singh H K, Ray T, Smith W. Surrogate assisted simulated annealing (SASA) for constrained multi-objective optimization[C]. Barcelona: Proceedings of the IEEE Congress on Evolutionary Computation, 2010.
[18]	Couckuyt I, Deschrijver D, Dhaene T. Fast calculation of multi-objective probability of improvement and expected improvement criteria for Pareto optimization[J]. Journal of Global Optimization, 2014, 60(3):575-594. doi: 10.1007/s10898-013-0118-2
[19]	Tian J, Tan Y, Zeng J, et al. Multi-objective infill criterion driven gaussian process-assisted particle swarm optimization of high-dimensional expensive problems[J]. IEEE Transactions on Evolutionary Computation, 2018, 23(3):459-472. doi: 10.1109/TEVC.2018.2869247
[20]	Fan C D, Hou B, Zheng J H, et al. A surrogate-assisted particle swarm optimization using ensemble learning forexpensive problems with small sample datasets[J]. Applied Soft Computing, 2020, 9(1):1-17. doi: 10.1016/j.asoc.2008.05.006
[21]	Wang H, Jin Y, Doherty J. Committee-based active learning for surrogate-assisted particle swarm optimization of expensive problems[J]. IEEE Transactions on Cybernetics, 2017, 47(9):2664-2677. doi: 10.1109/TCYB.2017.2710978 pmid: 28650832
[22]	Kennedy J, Eberhart R. Particle swarm optimization[C]. Perth:Proceedings of the ICNN'95-IEEE International Conference on Neural Networks,1995.
[23]	Krige D G. 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]	Hardy R L. Multi-quadric equations of topography and other irregular surfaces[J]. Journal of Geophysical Research, 1971, 76(8):1905-1915. doi: 10.1029/JB076i008p01905
[25]	Park J S. Optimal latin-hypercube designs for computer experiments[J]. Journal of Statistical Planning and Inference, 1994, 39(1):95-111. doi: 10.1016/0378-3758(94)90115-5
[26]	陈万芬, 王宇嘉, 林炜星. 异构集成代理辅助多目标粒子群优化算法[J]. 计算机工程与应用, 2021, 57(12):77-85.
	Chen Wanfen, Wang Yujia, Lin Weixin. Heterogeneous ensemble surrogate assisted multi objective particle swarm optimization algorithm[J]. Computer Engineering and Applications, 2021, 57(12):77-85.
[27]	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. doi: 10.1023/A:1008306431147
[28]	Deb K, Thiele L, Laumanns M, et al. Scalable multi-objective optimization test problems[C]. Honolulu: Proceedings of the Congress on Evolutionary Computation, 2002.
[29]	Sierra M R, Coello C A C. Improving PSO based multi objective optimization using crowding, mutation and ε-dominance evolutionary multi criterion optimization[C]. Heidelberg: Proceedings of the International Conference on Evolutionary Multi-Criterion Optimization, 2005.
[30]	Deb K, Jain H. An evolutionary many-objective optimization algorithm using reference-point-based non- dominated sorting approach, part I:Solving problems with box constraints[J]. IEEE Transactions on Evolutionary Computation, 2013, 18(4):577-601. doi: 10.1109/TEVC.2013.2281535
[31]	Van Veldhuizen D A, Lamont G B. Multi-objective evolutionary algorithm research:A history and analysis[J]. Evolutionary Computation, 1998, 8(2):125-147. doi: 10.1162/106365600568158
[32]	Schott J R. Fault tolerant design using single and multi-criteria genetic algorithm optimization[M]. Cambridge: Massachusetts Institute of Technology, 1995.
[33]	While L, Hingston P, Barone L, et al. A faster algorithm for calculating hyper-volume[J]. IEEE Transactions on Evolutionary Computation, 2006, 10(1):29-38. doi: 10.1109/TEVC.2005.851275

算法名称	代理模型方法	模型管理策略
OMOPSO^[29]	无	无
NSGAIII^[30]	无	无
HE-OMOPSO	异构集成模型	无
HE-OMOPSO-EI	异构集成模型	单一加点准则
HE-OMOPSO-CIP	异构集成模型	组合加点准则

测试函数	变量范围	函数特性
DTLZ1	D=7, 0≤x_i≤1	多模态,包含多个局部极值,难以收敛全局Pareto前沿
DTLZ2	D=12, 0≤x_i≤1	具有测试问题多样性维护的能力
DTLZ3	D=12, 0≤x_i≤1	多模态,包含多个局部极值,难以收敛全局Pareto前沿
DTLZ4	D=12, 0≤x_i≤1	DTLZ2具有偏差区域的变体,参数α=100
DTLZ7	D=12, 0≤x_i≤1	具有多个不连续的Pareto前沿

测试函数	种群规模	决策变量	真实函数评估次数	近似函数评估次数
DTLZ1	100	7	10 000	1 000
DTLZ2	100	12	10 000	1 000
DTLZ3	100	12	10 000	1 000
DTLZ4	100	12	10 000	1 000
DTLZ7	100	12	10 000	1 000

测试函数	M	D	指标	NSGAIII	OMOPSO	HE-OMOPSO	HE-OMOPSO-EI	HE-OMOPSO-CIP
DTLZ1	3	7	GD	0.781 783	1.637 838	0.015 461	0.013 791	0.016 580
DTLZ1	3	7	SP	1.057 316	0.719 976	1.012 225	0.979 615	1.165 889
DTLZ2	3	12	GD	0.004 032	0.003 235	0.007 579	0.010 306	0.020 607
DTLZ2	3	12	SP	0.077 598	0.064 806	0.092 176	0.083 077	0.098 543
DTLZ3	3	12	GD	4.753 833	1.074 708	0.061 030	0.061 110	0.047 881
DTLZ3	3	12	SP	2.132 415	2.581 264	3.035 824	3.051 903	2.937 612
DTLZ4	3	12	GD	0.033 529	0.047 561	9.313 025	3.448 13	1.682 354
DTLZ4	3	12	SP	0.090 833	0.090 203	0.127 381	0.429 518	0.421 106
DTLZ7	3	12	GD	0.364 110	0.450 572	0.476 337	0.750 161	1.530 544
DTLZ7	3	12	SP	0.124 177	0.081 720	0.139 025	0.263 528	0.229 165

测试函数	M	D	指标	NSGAIII	OMOPSO	HE-OMOPSO	HE-OMOPSO-EI	HE-OMOPSO-CIP
DTLZ1	3	7	GD	0.843 610	1.491 720	0.001 272	0.001 554	0.001 721
DTLZ1	3	7	SP	0.311 322	0.179 325	0.121 373	0.117 135	0.083 827
DTLZ2	3	12	GD	0.000 174	0.000 115	0.000 605	0.000 792	0.001 627
DTLZ2	3	12	SP	0.003 854	0.005 398	0.006 559	0.006 725	0.005 243
DTLZ3	3	12	GD	2.392 910	6.811 090	0.006 446	0.004 494	0.003 796
DTLZ3	3	12	SP	0.397 680	0.698 996	0.172 701	0.230 019	0.230 019
DTLZ4	3	12	GD	0.013 921	0.011 886	3.353 000	5.088 000	1.342 000
DTLZ4	3	12	SP	0.026 589	0.020 837	0.077 574	0.193 824	0.159 270
DTLZ7	3	12	GD	0.007 771	0.021 923	0.207 848	0.226 970	0.364 183
DTLZ7	3	12	SP	0.022 571	0.007 540	0.114 841	0.131 889	0.058 693

Surrogate Assisted Multi-Objective Particle Swarm Optimization Based on Combined Infill Sampling Criterion

RichHTML

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 11

References 33

Related Articles 1

Metrics

Comments

Recommended 10

测试函数	M	D	NSGAIII	OMOPSO	HE-OMOPSO	HE-OMOPSO-EI	HE-OMOPSO-CIP
DTLZ1	3	7	0.958 385	0.937 484	0.912 172	0.811 930	0.907 399
DTLZ2	3	12	0.977 874	0.960 337	0.381 782	0.513 731	0.850 874
DTLZ3	3	12	0.961 162	0.970 020	0.994 284	0.992 462	0.985 235
DTLZ4	3	12	0.911 239	0.957 262	0.958 300	1.000 000	1.000 000
DTLZ7	3	12	0.955 206	0.949 845	1.000 000	1.000 000	0.980 100

测试函数	M	D	NSGAIII	OMOPSO	HE-OMOPSO	HE-OMOPSO-EI	HE-OMOPSO-CIP
DTLZ1	3	7	0.037 100 00	0.014 773 0	0.006 699	0.018 626 000	0.009 946
DTLZ2	3	12	0.000 083 83	0.000 305 0	0.016 939	0.009 872 000	0.008 428
DTLZ3	3	12	0.060 860 00	0.005 867 0	0.001 181	0.001 548 000	0.002 919
DTLZ4	3	12	0.032 718 00	0.001 587 0	0.044 901	0.000 001 654	0.000 000
DTLZ7	3	12	0.013 348 00	0.002 035 1	0.004 829	0.012 341 000	0.036 405