CASIA OpenIR  > 毕业生  > 博士学位论文
基于多源知识的二型模糊系统设计及其控制方法研究
Alternative TitleStudy on Multi-source Knowledge Based Type-2 Fuzzy Logic Systems Design and Control Methods
王铁超
Subtype工学博士
Thesis Advisor易建强
2011-11-29
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline控制理论与控制工程
Keyword二型模糊逻辑系统 多源知识 最小二乘法 积极集法 遗传算法 Narx模型 逆控制 Type-2 Fuzzy Logic System Multi-source Knowledge Least Squares Algorithm Active-set Algorithm Genetic Algorithm Narx Model Inverse Control
Abstract近年来二型模糊系统及其控制方法越来越成为一大研究热点,但很多与二型模糊逻辑系统及控制方法相关的 理论和技术还不完善,需要研究和解决。作为一型模糊逻辑系统的扩展,二型模糊逻辑系统在处理不确定性、 抗干扰、减少模糊规则数目等方面具有明显优势。研究表明融合先验知识所得到的系统模型能更准确 更可靠地体现原系统,具有更好的泛化能力。如果能在二型模糊系统设计过程中充分地 利用包括先验知识在内的各种信息(如经验和数据),将有助于进一步提高所设计的二型模糊系统的性能。为此, 本课题结合国家自然科学基金面上项目“基于多源知识的二型模糊系统设计与控制研究 (60975060)”和863计划项目“实现载荷水平调节的四绳索垂直牵引装卸机器人研制”(2007AA04Z239), 深入探讨二型模糊系统设计的相关理论和控制问题,并把相关的理论成果应用与实际。本文的主要工作和贡献有: 1、针对单输入单输出零阶非归一化区间二型模糊逻辑系统提出并证明了确保 对称性、有界性、单调性和凸性融入该模糊系统的充分条件,给出了基于先验知识和训练数据 利用约束最小二乘算法实现该二型模糊系统优化设计的方法和步骤。 数值仿真验证提出的充分条件的有效性及方法的优越性。 2、针对单输入单输出一阶归一化区间二型模糊逻辑系统提出并证明了确保对称性、 单调性和特殊点性质融入到该模糊系统的充分条件,分别给出了基于先验知识和训练数据 利用约束最小二乘法、积极集法和混合学习算法 实现该模糊系统的优化设计的方法和步骤。数值仿真验证了提出的充分条件的有效性,比较了三种算法的差异。 3、针对多输入单输出非归一化区间二型模糊逻辑系统提出并证明了 确保对称性(关于原点的对称性、关于特殊平面的镜像对称性)和单调性与该模糊系统融合的充分条件, 分别给出了基于先验知识和训练数据利用约束最小二乘法和遗传算法优化设计该模糊系统的方法和步骤。 数值仿真验证了提出的充分条件的有效性,比较了两种优化算法的优缺点。 4、采用非归一化区间二型模糊NARX模型基于先验知识和训练数据实现了双容水箱液位系统的辨识,仿真验证了 融合先验知识的二型模糊系统设计方法的有效性和优越性。基于先验知识、经验知识和训练数据利用一般训练和专门训练方法 分别求得由非归一化模糊NARX模型构成的直接逆控制器和自适应直接逆控制器并应用于针对绳索牵引自动水平 调节吊具系统中,实验结果表明了基于多源知识的直接逆控制的有效性和优越性。
Other AbstractIn recent years, interest in type-2 subjects is worldwide and touches on a broad range of application and many theoretical topics. As an extension of type-1 fuzzy logic, type-2 fuzzy logic has obvious advantages for handling different sources of uncertainties, reducing the number of fuzzy rules, interference suppression, etc. Researches demonstrate that incorporating prior knowledge into grex-box models can better represent the modeled objective, and obtain better generalization performance. If different kinds of information are blended which include prior knowledge, experience of operators and designers and measurements in the process of modeling and identification, the performance of the modeling can be further improved. Under the support of the National Natural Science Foundations of China (60975060) and the 863 Program (2007AA04Z239), novel design methods and control schemes of type-2 fuzzy logic systems are developed. The main contributions of the thesis include the following issues: For single-input single-output zeroth order unnormalized interval type-2 fuzzy logic system some sufficient conditions are presented which ensure that the prior knowledge -- bounded range, symmetry, monotonicity and convexity can be combined with this type-2 fuzzy system. The steps is given by which the type-2 fuzzy system is optimally designed via constraint least squares algorithm based on the prior knowledge and data. Numerical simulations verify the effectiveness of the sufficient conditions and the superiority of the methods. For single-input single-output first order normalized interval type-2 fuzzy logic system some sufficient conditions are presented which ensure that the prior knowledge -- symmetry, monotonicity and the property of special points can be combined with this type-2 fuzzy system. The steps is given by which the type-2 fuzzy system is optimally designed via constraint least squares algorithm, active-set algorithm and their hybrid algorithm based on the prior knowledge and data. Numerical simulations verify the effectiveness of the sufficient conditions, and compare the difference of the three algorithms. For multi-input single-output zeroth order unnormalized interval type-2 fuzzy logic system some sufficient conditions are presented which ensure that the prior knowledge -- symmetry ( symmetry about origin or a special plane) and monotonicity can be combined with this type-2 fuzzy system. The steps is given by which the type-2 fuzzy syst...
shelfnumXWLW1711
Other Identifier200918014628013
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/6402
Collection毕业生_博士学位论文
Recommended Citation
GB/T 7714
王铁超. 基于多源知识的二型模糊系统设计及其控制方法研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2011.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20091801462801(6092KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[王铁超]'s Articles
Baidu academic
Similar articles in Baidu academic
[王铁超]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[王铁超]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.