【作者】 齐波；

【导师】 吴凤祥；

【作者基本信息】 河北农业大学 ， 计算机应用技术， 2009， 硕士

【摘要】 落叶松根系为树形分枝结构,主要由初生根、次生根、初生根一次侧根、次生根一次侧根组成。各级根系之间具有明显的自相似性,本文基于L-系统理论构建落叶松根系生长模拟模型,表达落叶松根系构型的拓扑结构。根据根系生长的生理功能,改进了根系的向水性模型和向地性模型,探讨了基于人工生命方法的约束条件对落叶松根系形态结构以及生理功能的影响,将落叶松根系形态结构模型和生理功能模型相结合。实现了落叶松根系生长的动态模拟。落叶松根系属直根系,由四种不同类型的根组成,即主根、次生根、侧根和不定根。尽管微观上具有明显的相似性,但由于根系在生长过程中受土壤结构、湿度等各种因素的影响,使得根系的延伸过程表现出随机波动的不规则形态。文中以落中松生理生态理论为依据,建立了描述落叶松根系生长参数的数据结构,使得落叶松根系的动态生长模型和其形态发生模型相融合,并应用随机生成技术,抽象出落叶松根系形态及生长规律的规则描述,在数据结构中引入生长控制参数以实现动态可控性。设计了模型的解法,实现了便于计算机模拟的数值算法。设计开发了基于B/S结构的模拟软件,实现了落叶松根系生长及水分需求的计算机模拟。由于自然植物的生存、进化、生长受环境影响,表现为区域性和时空变异性。对模型的优化,引入了分种群的竞争模式,采用基于种群分类的进化算法。落叶松根系生长过程中会出现异常—即遗传算法的选择、变异和进化,将遗传算法的操作加入在分形过程中,对形态模型中L—系统所涉及到的参数通过遗传算法进行重新设计,可生成结构有不同变化的根系形态。对功能模型采用浮点数编码代替二进制编码,将性质相近的个体作为一个种群,初期进化搜索仅限种群内部,使个体进化不受其他种群的影响,每个种群均具有产生优秀个体的机会;进化计算几代后,进行所有种群个体的相互竞争和进化,促使较差个体在种群竞争中逐渐消亡,充分保持群体的多样性,有效地克服了过早收敛和局限于局部最优解的问题。对所用的进化算子进行了详细的描述,鉴于变异操作在浮点数编码算法中的重要性,本文采用自适应调整变异算子,根据个体的适应值调整变异的步伐,使较好的个体在较小的局部区域内小步伐地变异搜索,增加局部寻优的精细度,较差的个体则加大变异幅度,使之尽快跳出较差的区域,进入较有希望的区域。更多还原

【Abstract】 The larch root which is composed of the primary root,the secondary root,the lateral root of primary root,and the lateral root of secondary root,is a branching tree structure.As all the levels of the roots have significantly self-similarity,this article is mainly to build a simulation model root growth of larch based on the L-system theory. According to the physiological function of root growth,it improve the root system of the larch on the water-based model and the geotropism model. It explore the impact on the morphology structure and physiological function of larch root by constraint based on artificial life methods.It have achieved dynamic simulation of the larch root growth because of the combination of morphology structure model and physiological function model.Larch root belongs to the straight root,which is composed of four different types of the roots.It include main root,secondary root, lateral root and adventitious root. Despite the obvious micro-similarity of the larch root,the root is affected by soil structure, humidity and other factors on the process of growth.So it makes that the process of the extension of the root system show irregular shape in random fluctuation patterns. In the text,it has established data structure,which can descripe the growth parameters of the larch root, on the basis of physiological ecological theory of larch.At the same time, dynamic growth model of the larch is integrated with morphogenetic model.Applying the random generation technology,it has abstracted rule description of the morphology and growth pattern of the larch root.It has introduced growth control parameters in order to achieve dynamic control in the data structure. It has achieved numerical algorithm which is easy to simulate in the computer with solution of the model design.Also,it has also achieved the computer simulation of the the larch root growth and water needs,with the simulation software based on B / S structure.As the environment have an impact on the survival,evolution and growth of the natural plant,it perform regional and spatial variability.The introduction of competition pattern in the sub-population is to optimize the model,and it has used evolutionary algorithm based on population category. The process of the larch root appears unusual,that is the choice,variation and evolution of the genetic algorithm.When the genetic algorithm is joined in the operation of the process of fractal,it can redesign parameters involved in the morphology structure model based on L-system.Finally it can generate different root morphology in the structure. The floating-point encoding is used in the functional model instead of binary encoding.The algorithm takes the kin individuals as population,at the initial stages,the evolution searching only in the population inside so that individuals’ evolution can not be effect by the other populations and every population has chances to produce excellent individuals.After evolutionary computing several generations,this algorithm let all of the individuals compete and evolve each other so that the worse individuals gradually dying out of the individual competition.This algorithm fully maintains the diversity of populations,effectively conquers the premature convergence and local optimum solution.It describes evolution operators in detail.For the importance of variation operation in floating-point number coding,the paper used self-adapting adjusting variation operator,adjusted the variation steps based on individual self-adapting value,so that the better individuals vary in little steps in local region,enhance the precision,the worse individuals increase the steps to get out of the worse regions and enter into the prospective regions.更多还原