协同进化算法研究及应用
|
摘要
协同进化是近几年来群智能算法领域兴起的一个研究热点,中外学者都对其进行了广泛的研究。本文在充分研读文献的基础上,对协同进化的效率机制、信息交流方式、拓扑结构及应用领域等方面开展研究,并取得如下成果:
(1)提出了新的人工鱼粒子群混合协同优化算法。该算法充分利用了人工鱼群算法与粒子群算法的互补性,发挥了协同进化算法利于混合不同算法的特性。实验表明算法具有收敛速度较快、精度较高的特点;
(2)提出了两种新的信息共享拓扑结构:网络信息共享模型和多群多层信息共享模型。这两种模型和以往的模型相比,更能充分将不同搜索机制、不同搜索方法的算法融合于协同进化的机制之中,为问题的求解提供了新思路和新方法;
(3)提出了一种新的协同进化信息交流机制——二次协同的信息交流机制。该机制将协同进化的信息交流方式从种群间的信息交流扩展到个体基因组的信息交流,实现了协同信息共享的宏观与微观的结合;
(4)扩展了协同进化算法的应用领域。本文将协同进化应用于化工参数估计优化、约束优化、丁烯烷化过程优化、圆填充等问题的求解,为协同进化算法的应用开辟了新的方向
Co-evolution is a research focus in the field of swarm intelligence algorithmin recent years, and Chinese and foreign scholars have conducted extensive studieson it. Based on the full research literature, the efficiency mechanisms, andinformation-exchange ways, and topology structure, and applications ofco-evolution, were all studied in this paper, and got some results as follows.
(1) A new artificial fish hybridized particle swarm optimization algorithmwas proposed. The new algorithm makes use of the complementarities of artificialfish swarm algorithm and particle swarm optimization algorithm, allows full playto the co-evolutionary algorithm’s characteristics that is beneficial to mix thecharacteristics of different algorithms. The results of experiments showed that thenew algorithm convergence speed and had more precision characteristics;
(2) Two new models of information-sharing topology structures, networkinformation sharing model and multi-group multi-layer information sharing model,were proposed. Compared with previous models, these models have moreadvantages in combining the different search mechanisms and the different searchmethods with the mechanisms of co-evolutionary, and they provide a new idea andnew methods for solving problem;
(3) The second collaboration of information exchange mechanisms that is anew information exchange mechanisms based on co-evolutionary, was proposed.By turning information exchange only among populations into informationexchange among the individual genome, it brought about the combination of macroand micro in sharing co-evolutionary information;
(4) The application field of co-evolutionary algorithm was extended.Co-evolutionary algorithm was applied to chemical parameter estimationoptimization, constrained optimization, butene alkylation processoptimization, circle packing in this paper, and it developed a new directionfor the applications of co-evolutionary algorithm
(1)提出了新的人工鱼粒子群混合协同优化算法。该算法充分利用了人工鱼群算法与粒子群算法的互补性,发挥了协同进化算法利于混合不同算法的特性。实验表明算法具有收敛速度较快、精度较高的特点;
(2)提出了两种新的信息共享拓扑结构:网络信息共享模型和多群多层信息共享模型。这两种模型和以往的模型相比,更能充分将不同搜索机制、不同搜索方法的算法融合于协同进化的机制之中,为问题的求解提供了新思路和新方法;
(3)提出了一种新的协同进化信息交流机制——二次协同的信息交流机制。该机制将协同进化的信息交流方式从种群间的信息交流扩展到个体基因组的信息交流,实现了协同信息共享的宏观与微观的结合;
(4)扩展了协同进化算法的应用领域。本文将协同进化应用于化工参数估计优化、约束优化、丁烯烷化过程优化、圆填充等问题的求解,为协同进化算法的应用开辟了新的方向
Co-evolution is a research focus in the field of swarm intelligence algorithmin recent years, and Chinese and foreign scholars have conducted extensive studieson it. Based on the full research literature, the efficiency mechanisms, andinformation-exchange ways, and topology structure, and applications ofco-evolution, were all studied in this paper, and got some results as follows.
(1) A new artificial fish hybridized particle swarm optimization algorithmwas proposed. The new algorithm makes use of the complementarities of artificialfish swarm algorithm and particle swarm optimization algorithm, allows full playto the co-evolutionary algorithm’s characteristics that is beneficial to mix thecharacteristics of different algorithms. The results of experiments showed that thenew algorithm convergence speed and had more precision characteristics;
(2) Two new models of information-sharing topology structures, networkinformation sharing model and multi-group multi-layer information sharing model,were proposed. Compared with previous models, these models have moreadvantages in combining the different search mechanisms and the different searchmethods with the mechanisms of co-evolutionary, and they provide a new idea andnew methods for solving problem;
(3) The second collaboration of information exchange mechanisms that is anew information exchange mechanisms based on co-evolutionary, was proposed.By turning information exchange only among populations into informationexchange among the individual genome, it brought about the combination of macroand micro in sharing co-evolutionary information;
(4) The application field of co-evolutionary algorithm was extended.Co-evolutionary algorithm was applied to chemical parameter estimationoptimization, constrained optimization, butene alkylation processoptimization, circle packing in this paper, and it developed a new directionfor the applications of co-evolutionary algorithm
引文
[1] J.H.Holland. Adaptation in Natural and Artificial Systems[M]. Cambridge: MIT press, 1975.
[2] P. B. Grosso. Computer Simulations of Genetic Adaptation:Parallel SubcomponentInteraction in a Multi-locus Model[D]., Ann Arbor MI: University of Michigan,.1995:
[3] Husbands P, F.Mill. Simulated Co-evolution as the Mechanism for Emergent Planning andScheduling[A]. In:R.K.Belew and L.B.Booker(C),Proceedings of the Fourth InternationalConference on Genetic Algorithms.Morgar:Kaufmann,1991:264-270
[4] Zhao.Q.F. Ageneral framework for cooperative algorithms:a society model[J].Proc IECE98:57-62.
[5] Ben Niu,Yun-long Zhu,Xiao-xian He. Multi-population Cooperative Partical SwarmOptimization[A]. In:Proceedings of the 8th European conference on advances in artificiallife,ECAL 2005[C], UK:Canterbury, 2005:874–883.
[6] Potter M A,De Jong K A.Cooperative coevolution:an architecture for evolving coadaptedsubcomponents[J].Evolutionary Computation.2000,8(1):1-29
[7] Tanhk Min-Jea, Sun Byung-Chan.Coevolutionary optimization based on ecologicalcooperation[J].Journal of software.2001,4(2):114-122.
[8] Hong-yun Meng, Xiao-hua Zhang,San-yang Liu. A Co-evolutionary Particle SwarmOptimization-Based Mehtod for Multiobjective Optimazation[A]. In:Proceedings of the 18thAustralian Joint Conference on Artificial Intelligence,AI 2005[C],Sydney: 2005: 349–359.
[9] Renato A.Krohling, Frank Hoffmann, Leandro dos Santos Coelho. Co-evolutionary ParticleSwarm Optimization for Min-Max Problems using Gaussian Distribution[A].In:Proceedings ofthe 2004 congress on evolutionary computation[C],Portland:IEEE:2004,959-964.
[10] Renato A.Krohling, Leandro dos Santos Coelho. Coevolutionary Particle SwarmOptimization Using Gaussian Distribution for Solving Constrained OptimizationProblems[J].IEEE Transactions on Systems,Man and Cybernetics,Part B,2006: 1407-1416.
[11] Mohammed El-Abd, Mohamed Kamel. Information Exchange in Multiple CooperatingSwarms[A], In:Proceedings of the 2005 conference on Genetic and evolutionarycomputation[C],2005:269-270.
[12] J.J.Liang, P.N.Suganthan. Dynamic Multi-Swarm Particle Swarm Optimizer[A],In:Proceedings of 2005 IEEE conference on Swarm Intelligence Symposium[C],2005: 124-129.
[13]李爱国.多粒子群协同优化算法[J].复旦学报(自然科学版),2004,43(5):923-925.
[14]王俊年,申群太,沈洪远,年晓红.基于协同进化微粒群算法的神经网络自适应噪声消除系统[J].计算机工程与应用,2005,13:20-23.
[15]王俊年,申群太,沈洪远，周鲜成.基于多种群协同进化微粒群算法的径向基神经网络设计[J].控制理论与应用,2006,23(2):251-255.
[16]李菲菲,姚坤,刘希玉.一种多微粒群协同进化算法[J].计算机工程与应用,2007,43(22):44-46.
[17]屈百达,焦竹青,徐保国.多量子粒子群协同优化算法研究[J].计算机工程与应用,2008,44(7):72-74.
[18]牛奔，李丽，楚湘华.新型多群体协同进化粒子群优化算法[J],2009,45(3):28-34.
[19]李晓磊，邵之江，钱积新.一种基于动物自治体的寻优模式:鱼群算法[J].系统工程理论与实践,2002,22(11):32-38.
[20]李晓磊,路飞，田国会,等.组合优化问题的人工鱼群算法应用[J].山东大学学报:工学版,2004,34(5):64-67.
[21] Kennedy J,Eberhart R C.Particle swarm optimization[A] In:Proc.IEEE Int.Conf.on NeuralNetworks[C].Perth,WA,Australia:IEEE Service Center,1995:1942-1948.
[22] Jaekson,W.C.MeDowell,M.E.Simulated annealing with dynamic Perturbations: amethodology for optimization[A].In:IEEE Aerospace Applications Conference[C],1990:181-191.
[23]刁东宇,赵英凯.带双重变异算子的自适应粒子群优化算法[J].计算机工程与设计,2009,30(5):1186-1188.
[24]莫愿斌,陈德钊,胡上序.混沌粒子群算法及其在生化过程动态优化中的应用[J].化工学报,2006,57(9):2123-2127.
[25]张创业,莫愿斌.一种基于协同进化思想的人工鱼和粒子群混合优化算法[J].广西民族大学学报(自然科学版).2009,15(1):74-77.
[26]张梅凤,邵诚,甘勇,等.基于变异算子与模拟退火混合的人工鱼群优化算法[J].电子学报,2006,34(80):1381-1385.
[27]陈振兴,叶华,刘今.铯-铷-钒系低温硫酸催化剂上SO2氧化反应速率的机理解释[J].化学反应工程与工艺,2001,17(2):119-123.
[28]郑启富,陈德钊.优进遗传算法及其在化工数据处理中的应用[J].浙江大学学报(工学版),2003,37(3):303-313.
[29]莫愿斌,陈德钊,胡上序.高维化工数据共轭粒子群算法处理[J].计算机工程与设计,2007(06):1241-1244.
[30]方强,陈德钊.优进策略的差分进化算法及其化工应用[J].化工学报,2004,55(4):598-602.
[31]王凌,刘波.微粒群优化与调度算法[M].北京:清华大学出版社,2008:85-87.
[32]高丽丽,刘弘,李同喜.基于文化粒子群算法的约束优化问题求解[J].计算机工程,2008,34(5):179-181.
[33]甘敏,彭辉.一种新的自适应惩罚函数算法求解约束优化问题[J].信息与控制,2009,38(1):25-28.
[34] Adjiman CS,Dallwig S,Floudas CA,Netunaier A.A global optimization method,alpha BB,forgeneral twice differentiable constrained NLPs-I[J].Theoretical Advance,1998,22(9):1137-1158
[35]贺益君,陈德钊.连续约束蚁群优化算法的构建及其在丁烯烷化过程中的应用[J].化工学报,2005,56(9):1708-1713.
[36]黄海燕,顾幸生,刘漫丹.求解约束优化问题的文化算法研究[J].自动化学报,2007,33(10):1115-1120.
[37] Beardon,Alan F,Kenneth Stephenson.The Uniformization Theorem for CirclePackings.[J].Indiana Univ.Math,1990,(39):1383-1425.
[38] Collins,Charles R.Kenneth Stephenson.A Circle Packing Algorithm [J].Comput.Geom.Theory Appli,2003(25):233-256.
[2] P. B. Grosso. Computer Simulations of Genetic Adaptation:Parallel SubcomponentInteraction in a Multi-locus Model[D]., Ann Arbor MI: University of Michigan,.1995:
[3] Husbands P, F.Mill. Simulated Co-evolution as the Mechanism for Emergent Planning andScheduling[A]. In:R.K.Belew and L.B.Booker(C),Proceedings of the Fourth InternationalConference on Genetic Algorithms.Morgar:Kaufmann,1991:264-270
[4] Zhao.Q.F. Ageneral framework for cooperative algorithms:a society model[J].Proc IECE98:57-62.
[5] Ben Niu,Yun-long Zhu,Xiao-xian He. Multi-population Cooperative Partical SwarmOptimization[A]. In:Proceedings of the 8th European conference on advances in artificiallife,ECAL 2005[C], UK:Canterbury, 2005:874–883.
[6] Potter M A,De Jong K A.Cooperative coevolution:an architecture for evolving coadaptedsubcomponents[J].Evolutionary Computation.2000,8(1):1-29
[7] Tanhk Min-Jea, Sun Byung-Chan.Coevolutionary optimization based on ecologicalcooperation[J].Journal of software.2001,4(2):114-122.
[8] Hong-yun Meng, Xiao-hua Zhang,San-yang Liu. A Co-evolutionary Particle SwarmOptimization-Based Mehtod for Multiobjective Optimazation[A]. In:Proceedings of the 18thAustralian Joint Conference on Artificial Intelligence,AI 2005[C],Sydney: 2005: 349–359.
[9] Renato A.Krohling, Frank Hoffmann, Leandro dos Santos Coelho. Co-evolutionary ParticleSwarm Optimization for Min-Max Problems using Gaussian Distribution[A].In:Proceedings ofthe 2004 congress on evolutionary computation[C],Portland:IEEE:2004,959-964.
[10] Renato A.Krohling, Leandro dos Santos Coelho. Coevolutionary Particle SwarmOptimization Using Gaussian Distribution for Solving Constrained OptimizationProblems[J].IEEE Transactions on Systems,Man and Cybernetics,Part B,2006: 1407-1416.
[11] Mohammed El-Abd, Mohamed Kamel. Information Exchange in Multiple CooperatingSwarms[A], In:Proceedings of the 2005 conference on Genetic and evolutionarycomputation[C],2005:269-270.
[12] J.J.Liang, P.N.Suganthan. Dynamic Multi-Swarm Particle Swarm Optimizer[A],In:Proceedings of 2005 IEEE conference on Swarm Intelligence Symposium[C],2005: 124-129.
[13]李爱国.多粒子群协同优化算法[J].复旦学报(自然科学版),2004,43(5):923-925.
[14]王俊年,申群太,沈洪远,年晓红.基于协同进化微粒群算法的神经网络自适应噪声消除系统[J].计算机工程与应用,2005,13:20-23.
[15]王俊年,申群太,沈洪远，周鲜成.基于多种群协同进化微粒群算法的径向基神经网络设计[J].控制理论与应用,2006,23(2):251-255.
[16]李菲菲,姚坤,刘希玉.一种多微粒群协同进化算法[J].计算机工程与应用,2007,43(22):44-46.
[17]屈百达,焦竹青,徐保国.多量子粒子群协同优化算法研究[J].计算机工程与应用,2008,44(7):72-74.
[18]牛奔，李丽，楚湘华.新型多群体协同进化粒子群优化算法[J],2009,45(3):28-34.
[19]李晓磊，邵之江，钱积新.一种基于动物自治体的寻优模式:鱼群算法[J].系统工程理论与实践,2002,22(11):32-38.
[20]李晓磊,路飞，田国会,等.组合优化问题的人工鱼群算法应用[J].山东大学学报:工学版,2004,34(5):64-67.
[21] Kennedy J,Eberhart R C.Particle swarm optimization[A] In:Proc.IEEE Int.Conf.on NeuralNetworks[C].Perth,WA,Australia:IEEE Service Center,1995:1942-1948.
[22] Jaekson,W.C.MeDowell,M.E.Simulated annealing with dynamic Perturbations: amethodology for optimization[A].In:IEEE Aerospace Applications Conference[C],1990:181-191.
[23]刁东宇,赵英凯.带双重变异算子的自适应粒子群优化算法[J].计算机工程与设计,2009,30(5):1186-1188.
[24]莫愿斌,陈德钊,胡上序.混沌粒子群算法及其在生化过程动态优化中的应用[J].化工学报,2006,57(9):2123-2127.
[25]张创业,莫愿斌.一种基于协同进化思想的人工鱼和粒子群混合优化算法[J].广西民族大学学报(自然科学版).2009,15(1):74-77.
[26]张梅凤,邵诚,甘勇,等.基于变异算子与模拟退火混合的人工鱼群优化算法[J].电子学报,2006,34(80):1381-1385.
[27]陈振兴,叶华,刘今.铯-铷-钒系低温硫酸催化剂上SO2氧化反应速率的机理解释[J].化学反应工程与工艺,2001,17(2):119-123.
[28]郑启富,陈德钊.优进遗传算法及其在化工数据处理中的应用[J].浙江大学学报(工学版),2003,37(3):303-313.
[29]莫愿斌,陈德钊,胡上序.高维化工数据共轭粒子群算法处理[J].计算机工程与设计,2007(06):1241-1244.
[30]方强,陈德钊.优进策略的差分进化算法及其化工应用[J].化工学报,2004,55(4):598-602.
[31]王凌,刘波.微粒群优化与调度算法[M].北京:清华大学出版社,2008:85-87.
[32]高丽丽,刘弘,李同喜.基于文化粒子群算法的约束优化问题求解[J].计算机工程,2008,34(5):179-181.
[33]甘敏,彭辉.一种新的自适应惩罚函数算法求解约束优化问题[J].信息与控制,2009,38(1):25-28.
[34] Adjiman CS,Dallwig S,Floudas CA,Netunaier A.A global optimization method,alpha BB,forgeneral twice differentiable constrained NLPs-I[J].Theoretical Advance,1998,22(9):1137-1158
[35]贺益君,陈德钊.连续约束蚁群优化算法的构建及其在丁烯烷化过程中的应用[J].化工学报,2005,56(9):1708-1713.
[36]黄海燕,顾幸生,刘漫丹.求解约束优化问题的文化算法研究[J].自动化学报,2007,33(10):1115-1120.
[37] Beardon,Alan F,Kenneth Stephenson.The Uniformization Theorem for CirclePackings.[J].Indiana Univ.Math,1990,(39):1383-1425.
[38] Collins,Charles R.Kenneth Stephenson.A Circle Packing Algorithm [J].Comput.Geom.Theory Appli,2003(25):233-256.