Page 204 - 《软件学报》2026年第3期

丁炜超 等: 基于信息共享的改进双归档高维多目标进化算法 1167

从图 11 结果可以看出, SMS-EMOA 算法因其特殊的指标设计导致运行时间复杂度较高, C-TAEA 算法次之,
Two-Arch/IS、Two_Arch2、TriMOEATAR、MOEA/D 算法的表现相似, 而其余算法表现出更低的计算复杂度.
Two-Arch/IS 相较于 Two-Arch 需要更多的运行时间, 这是因为 Two-Arch/IS 引入了多种改进机制提升算法性能,
虽然带来一定的计算负担, 但根据前文的实验结果及分析, Two-Arch/IS 相较于 Two-Arch 实现了显著的性能提升.
尽管 Two-Arch/IS 在运行时间上并不占优势, 但与其他算法的差距处于合理范围内, 并未显示出明显劣势.

4 结 论

本文提出了一种面向 MaOP 的双归档算法 Two-Arch/IS, 能够有效缓解 MaOP 中收敛性和多样性冲突导致种
群难以收敛的困境. 为了应对算法在探索高维目标空间时面临的挑战, Two-Arch/IS 设计了一种限制性的繁殖策略,
通过选择亲本产生表现更佳的子代个体, 更有效地搜索解空间. 为了保证种群更有效地逼近 PF, Two-Arch/IS 重新
设计了档案库的环境选择机制, 增加一组权重向量用于划分目标空间, 进而通过子种群更新两个档案库, 增强了种
群的多样性表现, 避免种群陷入局部最优解, 同时借助角度选择与转移密度估计的方法移除档案库中劣势解, 在有
效增强种群选择压力的同时极大程度上维护了种群多样性. 为进一步加强档案库间信息交流, Two-Arch/IS 设计了
基于边界解的个体迁移机制, 通过边界解引导档案库的进化方向, 实现了 CA 和 DA 的优势互补. 通过在不同类型
的测试问题上进行性能测试, Two-Arch/IS 相比于其他代表性算法在 MaOP 上表现出较为明显的优势, 同时在低维
上也展现出良好的有效性. 未来的工作可以聚焦于探索更高效的个体迁移方法, 如根据当前档案库状态实现自适应
的个体迁移, 通过动态调整个体迁移策略及相关参数, 从而更精准的指导搜索过程, 增强算法的适应性与综合性能.

References
[1] Champasak P, Panagant N, Pholdee N, Bureerat S, Yildiz AR. Self-adaptive many-objective meta-heuristic based on decomposition for
many-objective conceptual design of a fixed wing unmanned aerial vehicle. Aerospace Science and Technology, 2020, 100: 105783. [doi:
10.1016/j.ast.2020.105783]
[2] Yao SG, Tian HQ, Xu P. Lightening design of carbody structure for heavy haul gondola. Journal of Traffic and Transportation
Engineering, 2011, 11(1): 31–35, 57 (in Chinese with English abstract). [doi: 10.19818/j.cnki.1671-1637.2011.01.006]
[3] Xie SC, Zhou H. Multi-objective optimisation of a vehicle energy absorption structure based on surrogate model. Journal of Central South
University, 2014, 21(6): 2539–2546. [doi: 10.1007/s11771-014-2209-8]
[4] Xiang Y, Yang XW, Zhou YR, Zheng ZB, Li MQ, Huang H. Going deeper with optimal software products selection using many-
objective optimization and satisfiability solvers. Empirical Software Engineering, 2020, 25(1): 591–626. [doi: 10.1007/s10664-019-
09761-2]
[5] Cai XJ, Zhang JJ, Ning ZH, Cui ZH, Chen JJ. A many-objective multistage optimization-based fuzzy decision-making model for coal
production prediction. IEEE Trans. on Fuzzy Systems, 2021, 29(12): 3665–3675. [doi: 10.1109/TFUZZ.2021.3089230]
[6] Li YH, Liu ZJ, Li J. A multi-objective optimization algorithm of switched reluctance generator based on fuzzy logic NSGA-III. Computer
Engineering & Science, 2021, 43(3): 534–541 (in Chinese with English abstract). [doi: 10.3969/j.issn.1007-130X.2021.03.020]
[7] Liu QQ, Jin YC, Heiderich M, Rodemann T, Yu G. An adaptive reference vector-guided evolutionary algorithm using growing neural gas
for many-objective optimization of irregular problems. IEEE Trans. on Cybernetics, 2022, 52(5): 2698–2711. [doi: 10.1109/TCYB.2020.
3020630]
[8] Yang Z, Dong L, Cai XJ. Toward cooperative multi-agent video streaming perception. SCIENTIA SINICA Informationis, 2023, 53(1):
46–65 (in Chinese with English abstract). [doi: 10.1360/SSI-2021-0179]
2
[9] Xie CW, Guo H, Wei W, Jiang L. MaOEA/d : Many-objective evolutionary algorithm based on double distances. Ruan Jian Xue
Bao/Journal of Software, 2023, 34(4): 1523–1542 (in Chinese with English abstract). http://www.jos.org.cn/1000-9825/6702.htm [doi: 10.
13328/j.cnki.jos.006702]
[10] Zou J, Yang QT, Yang SX, Zheng JH. Ra-dominance: A new dominance relationship for preference-based evolutionary multiobjective
optimization. Applied Soft Computing, 2020, 90: 106192. [doi: 10.1016/j.asoc.2020.106192]
[11] Liu Y, Zhu NB, Li KL, Li MQ, Zheng JH, Li KQ. An angle dominance criterion for evolutionary many-objective optimization.
Information Sciences, 2020, 509: 376–399. [doi: 10.1016/j.ins.2018.12.078]
[12] Li MQ, Yang SX, Liu XH. Shift-based density estimation for Pareto-based algorithms in many-objective optimization. IEEE Trans. on
Evolutionary Computation, 2014, 18(3): 348–365. [doi: 10.1109/TEVC.2013.2262178]