混合式遗传算法及其在指派问题与排序问题中的应用-应用数学专业论文.docxVIP

西南交通大学硕士研究生学位论文 西南交通大学硕士研究生学位论文 第J页 摘要 ／ ／遗传算法由美国教授J．Holland提出，它是一类起源于生物选择和遗传机 制的随机搜索算法。遗传算法的主要特点是群搜索策略以及个体间信息交换 和搜索不依赖梯度信息，它特别适合于传统方法难于解决的复杂和非线性问 题并广泛应用于机器学习、自适应控制、组合设计、人工智能等领域。因此， 遗传算法是二十一世纪智能计算的一种关键技术。、】 本论文的工作集中在以下四个方面： 】．详细讨论了遗传算法的理论以及运行框架； 2．提出了一种新的改进遗传算法一混合式遗传算法，设计了解决函数优 化混合式遗传算法运行程序。仿真结果表明混合式遗传算法与其它几种遗传 算法相比有更好的性能； 3．运用混合式遗传算法较好地解决了指派问题。特别地，提出了～种改 进的交叉算子--PMX并设计了执行程序： 4．提出了生产管理中一类排序问题的数学模型并设计了有效的混合式 遗传算法求解程序。 指派问题和排序问题是～类复杂的组合问题，传统的方法难于解决。相 比之下，遗传算法解决此类问题具有优势。(特别地，仿真结果表明混合式遗 传算法混合式遗传算法由于结合了其它几种遗传算法的优点具有更好的性 能。另外舟本文针对指派问题和排序问题提出的PMX交叉算子以及设计的 运行程序非常有效。 关键词：遗传算法；混合式遗传算法；函数优化；指派问题：排序阅题。 AbstractGenetic Abstract Genetic Algorithm(GA)，developed by American professor J．Holland,is a sort of randomly searching aI906thms which originated from the creature rule--Nature Selection and Genefc Mechanism,and has a main chmac把ristic which is that the group searching strategy and the switching and searching for message between individuals are independent ofgradient information．n especially suits to the sophisticated and no．1inear problems which ale di伍cult to the traditional methods and carl be widely applied in the yields of machine learning， adaptive controlling,combinatory designing，artificial life and SO 011．Thus,it is one ofthe key techniques for intelligent computing in the 21 st century． This thesis iS focused 0n the following four topics． 1．nle theory ofGA and its operating frdme are discussed in detail+ 2．A new type of improved Ga§—I移brid Genetic Algorithm(riGA)is put forward,and its program is designed tO solve the problem offunction optimization． The computational results indicate that the performance of HGA iS better than some other GAs’． 3．HGA is applied to solve the assignment problem quite well．Especiallg a sort of improved crossover opemtors----PMX is brought forward，and its responding compufer program is designed． 4．The mathemalical mode ofa sequencing problem in management system is raised and it is solved successfully by designing the responding HGA comput