基于多源信息融合的变压器故障诊断方法研究
文献类型：学位论文
| 作者 | 丁腾飞
|
| 学位类别 | 硕士 |
| 答辩日期 | 2016-05-31 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 导师 | 曹云侠 |
| 关键词 | 变压器 故障诊断 DGA 多源信息融合 |
| 其他题名 | Research on Transformer Fault Diagnosis Based on Multi-source Information Fusion |
| 学位专业 | 控制理论与控制工程 |
| 中文摘要 | 本文首先介绍大型油浸式电力变压器故障诊断的研究现状，分析变压器常用的故障诊断方法和不足之处。然后，阐述油浸式电力变压器的基本结构，变压器内部故障和故障特征之间的耦合关系。最后，详细介绍多源信息融合的故障诊断框架下故障诊断的两种具体方法。多源信息融合的故障诊断框架主要由基于DGA的故障诊断方法和基于电气及相关量的故障诊断方法组成，把变压器内部作为一个整体对待，避免只采用单一方法的局限性和只针对某一特征量诊断的片面性。基于DGA的故障诊断方法将变压器油中5种特征气体作为输入，采用ICSBP算法根据气体的组分和含量与不同故障之间的关系来诊断故障的类型和程度。ICSBP算法是改进的布谷鸟搜索算法优化BP神经网络，在布谷鸟搜索算法中引入惯性权重，同时在步长和发现概率方面做出改进。通过验证，ICSBP在收敛速度和分类准确性等方面都有提高，可以更好地应用于变压器故障诊断中。基于电气及相关量的故障诊断方法是由各类传感器、调理电路、AD转换电路和DSP等组成在线监测系统对变压器内部重点对象：铁芯、绕组、微水和油温等进行实时监测，可以及时发现DGA难以察觉的突发性异常。最后，把两种方法的故障诊断结果进行融合，对变压器的内部故障做出更加准确全面的诊断。 |
| 英文摘要 | At first, this paper introduced the research status of large oil-immersed transformer fault diagnosis and then analyzed the common diagnostic method for transformer and its shortcomings. Then this paper introduce the basic structure of oil-immersed power transformer and the relationship between internal fault of transformer and fault characteristics. Finally, this paper introduces two methods of fault diagnosis under the framework of multi-source information fusion in detail. Fault diagnosis framework of multi-source information fusion is composed of fault diagnosis method based on DGA and fault diagnosis method based on the electrical and related quantity. In this framework, it treats the internal transformer as a whole, which can avoid limitations and sidedness by a single method of diagnosis or only a particular characteristic. The input variable of fault diagnosis method based on DGA is the five types of gas, and the diagnosis principle is based on ICSBP algorithm, which can diagnose the fault type and degree according to the relationship between different characteristics of the gas (ie., composition &content) and the different faults. It is noted that ICSBP algorithm is the improved cuckoo search algorithm, which can be used to optimize BP neural network. In detail, this algorithm realized three improvements in terms of ω, Pa and ?. Also, the further validation proved ICSBP algorithm can do better than the previous one in terms of convergence speed and classification accuracy, thus it can be better applied to transformer fault diagnfosis. Besides, fault diagnosis method based on the electrical and related quantity is made up of various types of sensors, conditioning circuit, AD, DSP etc. The method can realize real-time monitoring for internal key objects of transformer, such as transformer core and winding, micro water, oil temperature, etc, which can find unexpected exception. The combination of this two fault diagnosis results can help make more accurate and comprehensive internal fault diagnosis for the transformer. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 页码 | 59页 |
| 源URL | [http://ir.sia.cn/handle/173321/19654]  |
| 专题 | 沈阳自动化研究所_信息服务与智能控制技术研究室 |
| 推荐引用方式 GB/T 7714 | 丁腾飞. 基于多源信息融合的变压器故障诊断方法研究[D]. 中国科学院沈阳自动化研究所. 2016. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。