论文总字数：37846字

摘 要

随着电力系统的不断扩大，对供电可靠性的要求日益提高，如何及时、有效地实现运行中的电力变压器故障诊断，已成为十分迫切和重要的任务。由于变压器故障征兆与故障原因间的关系存在着不确定性、复杂性，而在处理不确定性问题上，粗糙集方法可以计算得到约简值而不需要任何先验知识和附加信息，被证明特别适用于数据约简。然而在传统代数粗糙集中，只有那些能完全包含于某个概念的等价类才能被判断为确定属于决策概念，在实际问题中，由于噪声等因素的存在，完全包含于决策类的条件等价类概念较少。因此，实际问题中采用定量的概率包含关系来度量集合对于决策概念的隶属度是很有必要的。决策粗糙集模型是一种概率粗糙集模型，与其他概率型粗糙集理论不同的是，决策粗糙集模型中区分正域、负域、边界域的阈值是通过计算各决策的最小风险得到的确定值，由于将贝叶斯决策方法引入到阈值的判断中，使得决策粗糙集应用于实际时具有充分的理论依据。

本文采用决策粗糙集的方法建立变压器故障诊断模型，首先利用决策粗糙集方法对原始数据进行约简，然后对约简后得到的数据进行规则提取，形成精简的规则集。以此基础构建的粗糙集模型完全是由决策粗糙集理论的最终约简规则决定的，减少了工作量，学习速度大为提高。最后结合变压器故障样本数据进行仿真，结果表明了该模型具有较高的精度。

关键词：电力变压器 故障诊断 决策粗糙集 规则提取

Transformer fault diagnosis based on Decision-Theoretic rough

Abstract

With the continuous expansion of the power system, the reliability of the power supply increasing demand. How timely and effectively find the fault of power transformers, has become a very urgent and important task. Because of the relationship between fault symptoms and causes of the transformer fault are uncertainty and complexity, Deal with uncertainty, Rough set can calculate reduction in value without the need for any prior knowledge and additional information, However, in the traditional algebraic rough set, only those completely contained in a concept of equivalence classes are identified in order to be judged making concept, Under actual conditions, due to the presence of noise and other factors, fewer conditions completely contained in the concept of equivalence classes of decision classes. So, the probability of the actual problem in a quantitative measure of the set contains the membership relationship to the concept of the decision-making is necessary. In the decision-making rough set theory, approximations in the form of probability of expression, and other probabilistic rough set theory is different is that the decision-making rough set model to distinguish between positive field, negative field, the threshold boundary region is by minimizing the risk calculated for each decision determine the value obtained, since the Bayesian decision method is introduced to determine the threshold values, when applied to the actual decision-rough set theory has sufficient basis.

Rough set model built on this basis, First, reduce the original data, and extracting rules of the data after reduction extraction to form a streamlined set of rules.Construction of rough set model entirely based on the decision rules, reducing the workload and improving the learning speed. Finally, combined the sample data of the transformer fault to simulate, the results showed that the high fault tolerance and effectiveness of the algorithm.

Key Words: Power Transformer; Fault diagnosis; Decision-Theoretic rough; Rules extraction

目 录

摘　要 I

Abstract II

目 录 III

第一章 绪论 1

1.1课题研究背景及意义 1

1.1.1 研究背景 1

1.1.2 研究意义 1

1.2 国内变压器故障诊断的研究现状 2

1.3 油中溶解气体产生机理与变压器故障的相关性分析 4

1.3.1 油中气体产生机理 4

1.3.2 判断故障类型、性质 4

1.4 论文结构 6

第二章 课题相关的理论知识 8

2.1 粗糙集理论 8

2.1.1 粗糙集理论的概况 8

2.1.2 粗糙集方法的特点 8

2.1.3 粗糙集理论的基础知识 9

2.2 决策粗糙集理论 11

2.3基于粗糙集的规则提取方法 14

2.4 本章小结 15

第三章 基于决策粗糙集的变压器故障诊断模型 16

3.1设计流程 16

3.2 设计步骤 17

3.2.1 样本数据集 17

3.2.2 连续属性离散化 17

3.2.3 属性约简 20

3.2.4规则提取 20

3.3 仿真实验 20

3.4 本章小结 24

第四章 总结与展望 25

4.1 总结 25

4.2 展望 25

参考文献 27

致谢 29

请支付后下载全文，论文总字数：37846字