摘要：模糊系统是一种普通的建模结构。模糊系统的特点在于它提供了人类语言和电脑之间的交流通道，这一特点也被科学家和工程师所接受。在模糊系统建模中通常采用数据驱动技术，因为人类语言不足以描述整个系统，尤其是很复杂的系统。在数据驱动中，最重要的难题是数据采集。在实际工程中，建立模糊系统会遇到两个问题。一是数据采集，通常来讲，通过实验来收集数据，然而在有些情况下，每一次实验的运行是费钱费力的。在这种系统数据很少的情况下，可以选用由试验设计来策划。第二个难题与系统的复杂程度相关联，要求设计者保证准确性的前提下，提取尽量少的规则。
本文在查阅了大量文献的基础上，通过解决这两个问题，建立起了效果比较理想的模糊系统。实验设计的方法可以解决数据采集的问题，本毕业论文选取了均匀设计和拉丁超立方两种实验设计方法。MATLAB函数功能的实现可以控制规则的数量，提取尽量少的规则的问题。将实验设计与模糊控制相结合，就可以得到最优的模糊建模。进一步，通过比较拟合和模糊控制两种方法得到的模型，可以分析出多少的数据量更适用于模糊建模，这对于实际应用的数据分析有重要的意义。19773
关键字：模糊控制 实验设计 均匀设计 拉丁超立方 模糊规则
The Research of Fuzzy Control Theory Based on the Design of Experiment
Fuzzy system modeling is a common structure. The characteristic of the fuzzy system is that it provides the communication channel between human language and computer. This feature has been accepted by the scientists and engineers. Drive technology is usually used in fuzzy system modeling data, because human language is not enough to describe the whole system, especially a very complicated system. The most important problem in data driven is data collection. In the practical engineering, the establishment of fuzzy system has two questions. One is the data collection. Generally speaking, collecting data through the experiment in each experiment operation is expensive effort. In this system with small data, we can choose by the experimental design to plan. The second problem is associated with the complexity of the system. This requires the designer to ensure the accuracy of the premise to extract the rules as little as possible.
In this paper, on the basis of consulting a large number of literatures, by solving the two problems, we establish the ideal fuzzy system effect. Experimental design method can solve the problem of data acquisition. This paper selected the uniform design and Latin Hypercube Sampling two methods of experimental design. The realization of the MATLAB function can control the number of rules, extracting rules as less as possible. Combining experiment design and fuzzy control, you can get the optimal fuzzy modeling. Further, by comparing the fitting and the fuzzy control model of the two methods, we can analyze the amount of data that is more suitable for the fuzzy modeling. It has an important significance for the actual application of data analysis.
Key words:　 fuzzy control,  design of experiment, uniform design,  Latin Hypercube Sampling, fuzzy rule  
目录
1．绪论    1
1．1建模方法综述    1
1．2调研情况    1
2．实验设计介绍    3
2．1实验设计理论    3
2．1．1实验设计概述    3
2．2实验设计方法    4
2．2．1随机区组设计    4
2．2．2单因素实验设计    4
2．2．3正交实验设计    4
2．2．4均匀实验设计    5
2．2．5超拉丁方设计    6
2．3国外内现状及优缺点    7 MATLAB基于实验设计的模糊控制建模研究:http://www.751com.cn/zidonghua/lunwen_11279.html