Uniform Experimental Design with Constrained Region Based on Fruit Fly Algorithm

doi:10.16182/j.issn1004731x.joss.21-0673

Abstract

Abstract:

To solve the problems that existing two-phase differential evolutionary algorithms still have poor diversity of population distribution and weak local search ability in solving uniform designs in constrained experimental region, a new two-phase fruit fly optimization algorithm (ToPFOA) based on uniform experimental design is proposed. In the first stage, fruit fly search strategy combined with differential operator, K-means clustering and external document updating the centers of clusters is used todynamically improve distribution diversity of population in constrained region. In the second stage, a new fruit fly operator is designed to improve local search ability in constrained region. The experimental results show that ToPFOA is superior to ToPDE and ToPDEEDA in terms of solution quality and stability.

Key words: uniform experimental design, constrained region, evolutionary algorithm, fruit fly optimization algorithm, differential operator

CLC Number:

TP391.9

Jiawei Zhou, Xin Du, Youcong Ni, Hu Zhang, Hao Zhang, Haoran Ni, Feng Wang. Uniform Experimental Design with Constrained Region Based on Fruit Fly Algorithm[J]. Journal of System Simulation, 2023, 35(1): 110-122.

Figures/Tables 17

Fig. 1

Fig. 2

Fig. 3

Table 1

Table 2

Table 3

Fig. 4

Table 4

Fig. 5

Fig. 6

Fig. 7

Table 5

Fig. 8

Fig. 9

Fig. 10

Table 6

Table 7

References 20

1	Fisher R A. Design of Experiments[J]. Br Med J(S0959-8146), 1936, 1(3923): 554-554.
2	张晓江. 基于计算机仿真的控制系统新型稳定性判据 [J]. 系统仿真学报, 2008, 20(13), 3468
	-3471.Zhang Xiaojiang. New Stability Criterion of Control System Based on Computer Simulation[J]. Journal of System Simulation, 2008, 20(13): 3468-3471.
3	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 (S0305-215X), 2016, 52(14): 79-105.
4	董文永, 李元香. 演化仿真优化的并行实现及其应用[J]. 计算机学报, 2002, 25(11): 1236-1242.
	Dong Wenyong, Li Yuanxiang. Parallel Implementation and Application of Evolutionary Simulation Optimization [J].Chinese Journal of Computer Science, 2002, 25(11): 1236-1242.
5	Gan N, Gu J. Hybrid Meta-model-based Design Space Exploration Method for Expensive Problems[J].Structural and Multidisciplinary Optimization(S1615-1488), 2019, 59(3): 907-917.
6	于雅楠. 改善旋翼飞行体气动力参数的正交实验优化设计[J]. 系统仿真学报, 2016, 28(3): 174-181.
	Yu Yanan. Orthogonal Test Optimization Design for Improving Aerodynamic Parameters of Rotor Flight [J]. Journal of System Simulation, 2016, 28(3): 174-181.
7	Wu Z, Wang D, Okolo PN, et al. Efficient Space-filling and Near-Orthogonality Sequential Latin Hypercube for Computer Experiments[J]. Computer Methods in Applied Mechanics and Engineering (S0045-7825), 2017, 324: 348-365.
8	Chen R B, Hsu Y W, Hung Y, et al. Discrete Particle Swarm Optimization for Constructing Uniform Design on Irregular Regions[J]. Computational Statistics & Data Analysis (S0167-9473), 2014, 72: 282-297.
9	Long T, Li X, Shi R, et al. Gradient-free Trust-region-based Adaptive Response Surface Method for Expensive Aircraft Optimization[J]. AIAA Journal, American Institute of Aeronautics and Astronautics (S1533-385X), 2018, 56(2): 862-873.
10	Akbaba H, Ozder M. Optimization and Screening of Solid Lipid Nanoparticle Production for Gene Delivery by Factorial Design and Response Surface Methodology[J]. Experimental Biomedical Research (S2618-6454), 2021, 4(1): 23-37.
11	Viana F A. A Tutorial on Latin Hypercube Design of Experiments. Quality and Reliability Engineering International[J]. Wiley Online Library (S1099-1638), 2016, 32(5): 1975-1985.
12	Sun Y, Meng X, Long T, et al. A Fast Optimal Latin Hypercube Design Method using an Improved Translational Propagation Algorithm[J]. Engineering Optimization, Taylor & Francis (S0305-215X), 2020, 52(7): 1244-1260.
13	Long T, Wu D, Chen X, et al. A Deterministic Sequential Maximin Latin Hypercube Design Method using Successive Local Enumeration for Metamodel-Based Optimization[J]. Engineering Optimization, Taylor & Francis (S0305-215X), 2016, 48(6): 1019-1036.
14	Wang Y, Xu B, Sun G, et al. A Two-phase Differential Evolution for Uniform Designs in Constrained Experimental Domains[J]. IEEE Transactions on Evolutionary Computation (S1089-778X), 2017, 21(5): 665-680.
15	魏佳宁, 郝昊, 常衢通, 等. 一种基于演化算法的约束区域均匀仿真实验设计方法[J]. 系统仿真学报, 2021, 33(7), 1591-1599.
	Wei Jianing, Hao Hao, Chang Qutong, et al. A Uniform Experimental Design with Constrained Region based on Evolutionary Algorithm[J]. Journal of System Simulation, 2021, 33(7), 1591-1599.
16	Liang J J, Runarsson T P, Mezura-Montes E, et al. Problem Definitions and Evaluation Criteria for the CEC 2006 Special Session on Constrained Real-parameter Optimization[J]. Journal of Applied Mechanics(S0021-8936), 2006, 41(8): 8-31.
17	王凌, 郑晓龙. 果蝇优化算法研究进展[J]. 控制理论与应用, 2017(5): 557-563.
	Wang Ling, Zheng Xiaolong. Advances in Fruit Fly Optimization Algorithms[J]. Control Theory & Applications, 2017(5): 557-563.
18	Stokes Z, Mandal A, Wong WK. Using Differential Evolution to Design Optimal Experiments[J]. Chemometrics and Intelligent Laboratory Systems(S0169-7439), 2020, 199: 103955.
19	杨俊闯, 赵超. K-means聚类算法研究综述[J]. 计算机工程与应用, 2019, 55(23): 7-14, 63.
	Yang Junchuang, Zhao Chao. Survey on K-Means Clustering Algorithm[J]. Computer Engineering and Applications, 2019, 55(23): 7-14, 63.
20	Borkowski J J, Piepel G F. Uniform Designs for Highly Constrained Mixture Experiments[J]. Journal of Quality Technology, Taylor & Francis (S0022-4065), 2009, 41(1): 35-47.

测试案例	变量维度	线性不等式约束	非线性不等式约束	线性等式约束	非线性等式约束	可行域比例
G04	5	0	6	0	0	51.123%
G05	4	2	0	0	3	<10^-3
G09	7	0	4	0	0	0.512%
G18	9	0	13	0	0	<10^-3
G21	7	0	1	0	5	<10^-3
crash box	14	0	17	1	0	<10^-3

案例	MD	ToPDE	ToPDEEDA	ToPFOA	案例	MD	ToPDE	ToPDEEDA	ToPFOA
G04	mean	0.473 693 384	0.471 434 981	0.458 114 605	G18	mean	0.068 926 397	0.065 627 623	0.062 813 074
G04	var	0.000 398 391	0.000 260 904	0.000 202 883	G18	var	4.818 85×10^-5	2.142 64×10^-5	7.525 56×10^-6
G05	mean	0.171 966 517	0.147 647 930	0.038 632 823	G21	mean	0.547 417 590	0.535 802 409	0.461 587 216
G05	var	0.004 493 037	0.000 902 097	0.000 284 999	G21	var	0.002 415 406	0.002 045 331	0.001 496 419
G09	mean	0.410 380 148	0.409 266 020	0.387 838 683	crash box	mean	0.835 646 260	0.785 208 958	0.763 069 539
G09	var	0.000 181 279	0.000 247 658	0.000 109 891	crash box	var	0.004 479 293	0.002 254 649	0.001 140 548

案例	MD	ToPDE	ToPDEEDA	ToPFOA	案例	MD	ToPDE	ToPDEEDA	ToPFOA
G04	mean	0.643 796 957	0.593 209 62	0.644 603 436	G18	mean	0.078 534 486	0.070 202 872	0.069 702 378
G04	var	0.001 749 655	0.002 226 158	0.002 378 382	G18	var	3.714 22×10^-5	2.203 93×10^-5	1.759 29×10^-5
G05	mean	0.266 575 712	0.178 637 771	0.042 720 925	G21	mean	0.761 078 093	0.872 306 329	0.774 844 110
G05	var	0.005 125 509	0.003 422 098	0.000 385 662	G21	var	0.017 847 936	0.005 599 352	0.010 898 765
G09	mean	0.524 948 193	0.476 324 503	0.490 785 918	crash box	mean	1.127 537 829	1.115 430 175	0.876 021 045
G09	var	0.000 516 334	0.000 352 793	0.001 205 851	crash box	var	0.014 386 173	0.009 923 292	0.003 768 711

案例	MD	ToPDE	ToPDEEDA	ToPFOA	案例	MD	ToPDE	ToPDEEDA	ToPFOA
G04	mean	0.482 785 304	0.473 803 624	0.447 619 323	G18	mean	0.068 649 945	0.070 952 918	0.063 975 721
G04	var	0.000 857 204	0.000 330 021	0.000 156 554	G18	var	2.114 48×10^-5	5.897 11×10^-5	2.207 44×10^-5
G05	mean	0.146 174 082	0.144 753 612	0.146 234 329	G21	mean	0.561 665 230	0.535 954 115	0.484 238 243
G05	var	0.001 217 454	0.001 143 475	0.001 219 613	G21	var	0.003 735 864	0.002 675 353	0.001 070 444
G09	mean	0.416 439 817	0.407 252 542	0.384 863 123	crash box	mean	0.859 704 033	0.763 296 718	0.778 463 458
G09	var	0.000 392 081	0.000 126 188	6.480 14×10^-5	crash box	var	0.003 377 056	0.002 949 788	0.002 439 232

案例	MD	ToPFOA_DE-	ToPFOA	案例	MD	ToPFOA_DE-	ToPFOA
G04	mean	0.648 177 836	0.644 603 436	G18	mean		0.069 702 378
G04	var	0.004 919 129	0.002 378 382	G18	var		1.759 29×10^-5
G05	mean		0.042 720 925	G21	mean	0.777 689 164	0.774 844 110
G05	var		0.000 385 662	G21	var	0.012 943 032	0.010 898 765
G09	mean	0.543 288 890	0.490 785 918	crash box	mean		0.876 021 045
G09	var	0.002 236 484	0.001 205 851	crash box	var		0.003 768 711