基于多变异策略与拥挤积距的多目标优化算法
|
摘要
为了进一步提升进化迭代中的群体多样性和分布性,提出一种基于多变异策略与拥挤积距的多目标优化算法(mcMOA)。该算法设计模糊记忆变异算子采集和利用进化中成功变异的尺度信息,以引导后续变异,增强了局部搜索效率。算法采用多变异策略,将模糊记忆变异、多项式变异、非一致性变异3种变异方式有机融入整个进化周期,提升了进化种群的多样性和全局搜索效率。针对拥挤距离不能有效表达个体局部分布性的问题,算法采用个体与相邻个体之间的距离乘积来替代拥挤距离,拥挤积距能有效表示个体的局部密度和局部分布性。通过标准测试函数的仿真实验并与多个采用单变异策略的经典算法比较,新算法在收敛性和分布性方面表现更优。
In order to further enhance the diversity and distribution of populations in evolutionary iterations,a multi-objective optimization algorithm based on multi-mutation strategy and crowded product distance is proposed. The algo-rithm designs fuzzy memory mutation operators to collect and use the scale information of successful mutations inevolution to guide subsequent mutations and enhance the efficiency of local search. The algorithm adopts multiplemutation strategies,including three kinds of fuzzy memory mutations,polynomial mutations,and non-consistentmutations. The variation method is organically integrated into the entire evolutionary cycle,which improves the di-versity of the evolutionary population and the global search efficiency. For the problem that crowding distance can-not effectively express the individual local distribution,the algorithm uses the distance product between the individ-ual and the adjacent individual to replace the crowded distance. The crowded product distance can effectively re-present the local density and local distribution of the individual. Through the simulation test of the standard testfunction and compared with a number of classic algorithms using a single mutation strategy,the new algorithm performs better in convergence and distribution.
In order to further enhance the diversity and distribution of populations in evolutionary iterations,a multi-objective optimization algorithm based on multi-mutation strategy and crowded product distance is proposed. The algo-rithm designs fuzzy memory mutation operators to collect and use the scale information of successful mutations inevolution to guide subsequent mutations and enhance the efficiency of local search. The algorithm adopts multiplemutation strategies,including three kinds of fuzzy memory mutations,polynomial mutations,and non-consistentmutations. The variation method is organically integrated into the entire evolutionary cycle,which improves the di-versity of the evolutionary population and the global search efficiency. For the problem that crowding distance can-not effectively express the individual local distribution,the algorithm uses the distance product between the individ-ual and the adjacent individual to replace the crowded distance. The crowded product distance can effectively re-present the local density and local distribution of the individual. Through the simulation test of the standard testfunction and compared with a number of classic algorithms using a single mutation strategy,the new algorithm performs better in convergence and distribution.
引文
[1] Deb K,Pratap A,Agrawal S,et al. A fast elitist multi-objective genetic algorithm:NSGA2[J]. IEEE Transac-tions on Evolutionary Computation,2002,6(2):182-197
[2] Zitzler E,Laumanns M,Thiele L. SPEA2:improving thestrength Pareto evolutionary algorithm[C]. In:Evolu-tionary Methods for Design,Optimization and Control withApplications to Industrial Problems. Athens:Internation-al Center for Numerical Methods in Engineering,2002.95-100
[3] Cai Z,Wang Y. A multiobjective optimization based evo-lutionary algorithm for constrained optimization[J]. IEEETransactions on Evolutionary Computation, 2006, 10(6):658-675
[4] Gong M G,Jiao L C,Du H F,et al. Multiobjective im-mune algorithm with nondominated neighbor-based selec-tion[J]. Evolutionary Computation,2008,16(2):225-255
[5] Antonelli M,Bernardo D,Hagras H,et al. Multiobjec-tive evolutionary optimization of type-2 fuzzy rule-basedsystems for financial data classification[J]. IEEE Trans-actions on Fuzzy Systems,2017,25(2):249-264
[6] Cheng R,Rodemann T,Fischer M,et al. Evolutionarymany-objective optimization of hybrid electric vehicle con-trol:from general optimization to preference articulation[J]. IEEE Transactions on Emerging Topics in Computa-tional Intelligence,2017,1(2):97-111
[7] Zuo Y,Gong M,Zeng J,et al. Personalized recommen-dation based on evolutionary multi-objective optimization[J]. IEEE Computational Intelligence Magazine,2015,10(1):52-62
[8] Gong M,Li H,Luo E,et al. A multi-objective coopera-tive coevolutionary algorithm for hyperspectral sparse un-mixing[J]. IEEE Transactions on Evolutionary Computa-tion,2017,21(2):234-248
[9] Du J,Jiang C,Chen K C,et al. Community-structuredevolutionary game for privacy protection in social networks[J]. IEEE Transactions on Information Forensics and Se-curity,2018,13(3):574-589
[10]高维尚,邵诚,高琴.群体智能优化中的虚拟碰撞:雨林算法[J].物理学报,2013,62(19):28-43
[11] Mininno E,Neri F,Cupertino F,et al. Compact differ-ential evolution[J]. IEEE Transactions on EvolutionaryComputation,2011,15(1):32-54
[12] Sabar N R,Ayob M,Kendall G,et al. Grammatical evo-lution hyper-heuristic for combinatorial optimization prob-lems[J]. IEEE Transactions on Evolutionary Computa-tion,2013,17(6):840-861
[13] Bouaziz S,Alimi A M,Abraham A. PSO-based updatememory for improved harmony search algorithm to the e-volution of FBBFNT’parameters[C]. In:Proceedings ofIEEE Congress on Evolutionary Computation(CEC),Beijing,China,2014. 1951-1958
[14] Zitzler E,Thiele L. Multi-objective evolutionary algo-rithms:a comparative case study and the strength Paretoapproach[J]. IEEE Transactions on Evolutionary Compu-tation,1999,3(4):257-271
[15]邓泽林,谭政冠,何锫,等.一种基于动态识别邻域的免疫网络分类算法及其性能分析[J].电子与信息学报,2015,37(5):1167-1172
[16] Qiu X,Xu J,Tan K C,et al. Adaptive cross-generationdifferential evolution operators for multi-objective optimi-zation[J]. IEEE Transactions on Evolutionary Computa-tion,2016,20(2):232-244
[17]段海滨,杨之元.基于柯西变异鸽群优化的大型民用飞机滚动时域控制[J].中国科学:技术科学,2018,48(3):277-288
[18]李雪岩,李雪梅,李学伟,等.基于混沌映射的元胞遗传算法[J].模式识别与人工智能,2015,28(01):42-49
[19] Hu W,Yen G G,Luo G. Many-objective particle swarmoptimization using two-stage strategy and parallel cellcoor-dinate system[J]. IEEE Trans on Cybernetics,2016,47(6):1446-1459
[20] Kumar R S,Kondapaneni K,Dixit V,et al. Multi-objec-tive modeling of production and pollution routingproblemwith time window:A self-learning particleswarm optimiza-tion approach[J]. Computers&Industrial Engineering,2016,99(C):29-40
[21]魏文红,王甲海,陶铭,等.基于泛化反向学习的多目标约束差分进化算法[J].计算机研究与发展,2016,53(6):1410-1421
[2] Zitzler E,Laumanns M,Thiele L. SPEA2:improving thestrength Pareto evolutionary algorithm[C]. In:Evolu-tionary Methods for Design,Optimization and Control withApplications to Industrial Problems. Athens:Internation-al Center for Numerical Methods in Engineering,2002.95-100
[3] Cai Z,Wang Y. A multiobjective optimization based evo-lutionary algorithm for constrained optimization[J]. IEEETransactions on Evolutionary Computation, 2006, 10(6):658-675
[4] Gong M G,Jiao L C,Du H F,et al. Multiobjective im-mune algorithm with nondominated neighbor-based selec-tion[J]. Evolutionary Computation,2008,16(2):225-255
[5] Antonelli M,Bernardo D,Hagras H,et al. Multiobjec-tive evolutionary optimization of type-2 fuzzy rule-basedsystems for financial data classification[J]. IEEE Trans-actions on Fuzzy Systems,2017,25(2):249-264
[6] Cheng R,Rodemann T,Fischer M,et al. Evolutionarymany-objective optimization of hybrid electric vehicle con-trol:from general optimization to preference articulation[J]. IEEE Transactions on Emerging Topics in Computa-tional Intelligence,2017,1(2):97-111
[7] Zuo Y,Gong M,Zeng J,et al. Personalized recommen-dation based on evolutionary multi-objective optimization[J]. IEEE Computational Intelligence Magazine,2015,10(1):52-62
[8] Gong M,Li H,Luo E,et al. A multi-objective coopera-tive coevolutionary algorithm for hyperspectral sparse un-mixing[J]. IEEE Transactions on Evolutionary Computa-tion,2017,21(2):234-248
[9] Du J,Jiang C,Chen K C,et al. Community-structuredevolutionary game for privacy protection in social networks[J]. IEEE Transactions on Information Forensics and Se-curity,2018,13(3):574-589
[10]高维尚,邵诚,高琴.群体智能优化中的虚拟碰撞:雨林算法[J].物理学报,2013,62(19):28-43
[11] Mininno E,Neri F,Cupertino F,et al. Compact differ-ential evolution[J]. IEEE Transactions on EvolutionaryComputation,2011,15(1):32-54
[12] Sabar N R,Ayob M,Kendall G,et al. Grammatical evo-lution hyper-heuristic for combinatorial optimization prob-lems[J]. IEEE Transactions on Evolutionary Computa-tion,2013,17(6):840-861
[13] Bouaziz S,Alimi A M,Abraham A. PSO-based updatememory for improved harmony search algorithm to the e-volution of FBBFNT’parameters[C]. In:Proceedings ofIEEE Congress on Evolutionary Computation(CEC),Beijing,China,2014. 1951-1958
[14] Zitzler E,Thiele L. Multi-objective evolutionary algo-rithms:a comparative case study and the strength Paretoapproach[J]. IEEE Transactions on Evolutionary Compu-tation,1999,3(4):257-271
[15]邓泽林,谭政冠,何锫,等.一种基于动态识别邻域的免疫网络分类算法及其性能分析[J].电子与信息学报,2015,37(5):1167-1172
[16] Qiu X,Xu J,Tan K C,et al. Adaptive cross-generationdifferential evolution operators for multi-objective optimi-zation[J]. IEEE Transactions on Evolutionary Computa-tion,2016,20(2):232-244
[17]段海滨,杨之元.基于柯西变异鸽群优化的大型民用飞机滚动时域控制[J].中国科学:技术科学,2018,48(3):277-288
[18]李雪岩,李雪梅,李学伟,等.基于混沌映射的元胞遗传算法[J].模式识别与人工智能,2015,28(01):42-49
[19] Hu W,Yen G G,Luo G. Many-objective particle swarmoptimization using two-stage strategy and parallel cellcoor-dinate system[J]. IEEE Trans on Cybernetics,2016,47(6):1446-1459
[20] Kumar R S,Kondapaneni K,Dixit V,et al. Multi-objec-tive modeling of production and pollution routingproblemwith time window:A self-learning particleswarm optimiza-tion approach[J]. Computers&Industrial Engineering,2016,99(C):29-40
[21]魏文红,王甲海,陶铭,等.基于泛化反向学习的多目标约束差分进化算法[J].计算机研究与发展,2016,53(6):1410-1421
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |