面向对象技术在沉积相建模中的应用研究

【作者】 张利锋；

【导师】 李从信；

【作者基本信息】 大庆石油学院 ， 计算机应用技术， 2004， 硕士

【摘要】 本文旨在建立一个复杂的系统—沉积相建模系统的00模型。随着软件规模的日益增长，系统的复杂性急剧上升。面向对象技术提供了驾驭系统复杂性的能力和手段。面向对象技术的关键是建模，建立正确的系统模型成为理解系统责任、乃至系统开发成功的关键。系统中使用标准建模语言UML对系统建模，UML具有很强的信息表达能力，但使用起来略感复杂，应根据需要进行裁剪。系统中采用了活动图、用例图、顺序图、交互图和类图建立沉积相建模系统的分析和设计模型。 建立模型需要过程指导，过程为软件开发的各个阶段提供了启发性规则和阶段成果建议。本质上说，00建模可以使用任何过程，我们采用UML作者推荐的统一过程RUP。RUP是一个通用的软件开发过程框架，它支持基于用例的、迭代增量式的开发，适合于小组多阶段的滚动开发。不同的阶段选择不同的用例，从而减低系统复杂性和开发的风险。RUP过程鼓励复用，设计过程中使用很多设计模式来增加复用性，如抽象工厂、观察者模式和迭代器模式等。模式可以使得软件结构稳定合理并具有较强的扩展性，同时具有较好的可维护性。 沉积相建模采用以地质统计学为基础的蒙特卡罗随机模拟方法。该法能够综合测井、录井、地震等多种数据，提供沉积相及油藏属性的多个可选择的、等概率的空间分布图像，这些图像的差异就反映了它们在空间分布的非均值性和不确定性。分析评价这种不确定性，能够为开发人员进行生产的分析和决策提供客观、定量的分析依据，有效降低油田开发的风险。系统给出了沉积相随机建模的分析类图和设计类图，实现了基于指示克里金估计的序贯指示随机模拟算法，该算法对离散数据和连续数据都适用，不要求数据的分布特征，可以较好的模拟相特征。研究了将遗传算法和模拟退火结合来增强算法全局寻优能力和运算效率，并用来模拟沉积属性参数的可能性，该算法是在标准遗传算法中融入局部搜索算法的思想，其特点主要体现在它引入了局部搜索的过程。基于群体中各个个体所对应的表现型，进行局部搜索，从而找出各个个体在目前环境下的局部最优解，以便达到改善群体总体性能的目的。最后给出了一个算法运行效率地比较结果。更多还原

【Abstract】 The thesis aims to build the object oriented model of a complicated system-sedimentary facies modeling system. The complexity ascending rapidly with the increasing of software scale. The object oriented technique provides the capability and methods to control the complexity .The key of the technique is modeling, the accurate model is the sticking point to comprehend the responsibilities of the software so much as which lead to the success of software devolopment. The system modeling use the standard modeling language UML, which provides strong power to express information but somewhat complicated ,so it needs customize. The activity diagram, use case diagram, sequence diagram, collaboration diagram and the class diagram are used for building the analysis and design model of the sedimentary facies modeling system..Build the modeling need process guidance, the process offers the illuminated rule and phase advice on modeling result. In essence, the OO modeling can use any process, The unifited process (RUP)which is recommend by the author of UML is adopted .RUP is a universal process framework for software development, it is based on the use case, iterative increment devolopment and adapt to the team multi-phase development. The different phase choose the special use cases, which decrease the complexity and the risk of development. Design patterns are also used to increase the reusability as RUP advocated .The abstract factory, observer and iterator patterns is introduced. The design patterns make the architecture of software reasonable, extendable and more maintainable.Sedimentary facies modeling use the Monte Carlo stochastic simulation method based on geostatistics. It can combine the seismic ,well logging and mud logging data source and provide multi-choosable and equality probability spatial images, the different images show the heterogeneity and uncertainty.Analysing and valuing the uncertainty offers objective and quantitative assessment standard to the oil engineer so as to reduce the risk of oil field production. The analysis and design class diagram are given. The sequence indication stochastic simulation algorithm based on indicator kriging is realized. The algorithm applies to discrete and continuous variable has no restrict to data distribution. Combination generic algorithm with simulated annealing to boost global optimization and efficiency is studied, the probability to simulate the sedimentary property is also taken into account. The algorithm combine the local search ideal with the standard generic algorithm, the main character is the local searching process is introduced. In order to improve the population performance, the individual local searching is used to get the local optimization on the environment. Finally, the comparison of algorithm efficiency is given.更多还原