【作者】 严小卫；

【导师】 樊明武；

【作者基本信息】 中国原子能科学研究院 ， 核技术及应用， 2001， 博士

【摘要】 加速器高频腔的设计对提高粒子加速器的性能有着十分重要的意义。而数据库技术作为计算机和信息科学技术发展最快的技术之一，自60年代起就被广泛用于各个领域。二者的结合已成为必然，对其研究已受到相当地重视。论文在如何利用计算机技术来辅助设计加速器腔这方面做了一些工作。 采用腔设计软件包来辅助设计加速器的高频腔，无疑能给我们带来较大的效益。而该软件包的设计与实现应属一个较复杂的软件工程，包括计算可视化、友好的用户界面、前／后处理等等。论文在软件包的前处理模块上做了一些实现工作，利用计算机图形处理技术开发了一个可视化的高频腔轮廓的辅助设计工具。借助该工具，我们可以在屏幕上绘出腔的轮廓，包括修改和编辑，然后程序自动产生所需的高频腔的定义文件，作为AUTOMESH的输入，改善了SUPERFISH的用户界面。该软件已实际应用，并产生了一定的经济效益。 为了利用数据库技术来管理和处理加速器腔设计时所需要的和产生的大量数据，论文设计和建立了一个加速器高频腔设计专用数据库管理系统原型FISHBASE，并初步实现了该系统原型的框架。该系统包括接口语言、词法和语法分析器、命令处理程序三层。 腔体电磁场的计算是设计一个加速腔的关键技术之一。问题之一便是给定一个腔的特征，如何找到一个具有该特征的较好的腔结构。论文采用随机优化方法来进行求解，提出了一种用于加速器高频腔设计的自适应遗传算法，并用FORTRAN语言进行了实现及验证。目录 中国原子能科学研究院博士学位论文 h 在加速器腔体的设计中，所面临的对象是复杂的，数据是时变的，所以我们有必要提出一个面向对象的、时态的数据模型。因此，论文提出了一个概念性数据模型，即时态实体一关系数据模型（TER模型厂提出了一个表示性数扼模型，即GT数据模型及其查询语言。并在理论上深入研究了基于TER模型的时态关系代数及类 SQL语言，为我们的加速器高频腔设计系统打下了一个良好的理论基础c 论文建立的加速器高频腔设计专用＊BMS原型＊＊BASE具有自己的特色，特别是其查询与传统数据库中的查询意义不一样，操作完全小同c论文给出的用于加速器高频腔设计的自适应遗传算洁＊＊G在数据结构和算法设计上也与经典遗传算法不同，具有自动调整基因以保证不产生异类的功能。TER模型及其代数系统的提出也是本论文工作的创新之一。更多还原

【Abstract】 Design of accelerator cavity is very important for the performance enhancement of particle accelerator. And as one of the most quickly developed techniques in the computer and information science, database technology has been applied to various areas since 60’s. Their combination becomes a necessity. And its studies turn to be a focus in both academic community and industrial community. We’ve devoted ourselves to the computer-aided design of accelerator cavity for some work.There is no doubt that we’ll much benefit from the usage of cavity design package to assist the design of RF cavity in accelerator. Further more, design and implementation of the package should be a rather complex software engineering, including computation visualization, friendly user interface, preprocess/postprocess, and so on. We do some implementation on the preprocess module, developing a visual tool to assist the design of RF cavity outline. By using this tool, we can visually draw cavity contour on the screen, modify and edit draft, and automatically generate the desired definition file of RF cavity, which is then used as the input of AUTOMESH module. The tool improves the user interface of SUPERFISH. This software has been applied in reality, producing certain economic benefits.In order to manage and handle the large amount of data which are needed and produced during the design of accelerator cavity, by applying the database technology, we design and establish a special database management system prototype for accelerator RF cavity design, called FISFDBASE. We also basically implement a framework of the system prototype. The system consists of three layers, i.e. interface language, lexical/syntactic parser, and command processor.Computation of electromagnetic field in cavity is one of the key techniques to design an accelerator cavity. One of the problems is that how we can find out a cavity skeleton which satisfies some given features. W7e use the random optimization method to solve this problem, proposing an adaptive genetic algorithm to design accelerator RF cavity. We also implement and verify this algorithm in FORTRAN language.During the design of accelerator cavity, we must face complicatedobjects and time-varied data. It is necessary for us to consider an object-oriented and temporal data model. Therefore, we give a conceptual data model, that is entity-relationship data model (shorten as TER model). At the same time, we propose a representative data model, i.e. GT data model, and its query language. Theoretically, we deeply study the temporal relational algebra and SQL-like language based on our TER model, laying a solid theoretic foundation for our accelerator RJF cavity design system.The special database management system prototype for accelerator RF cavity design, i.e. FISHBASE, has its own characteristics. Especially, its query is different from that in traditional databases, and their operations are totally unlike. The adaptive genetic algorithm for design of accelerator RF cavity, i.e. SAG, is also not the same as the classical genetic algorithm on both data structure and algorithm design. SAG can automatically adjust the genes in order not to generating unwonted individual. TER model and its algebraic system are one of our creative work too.更多还原