【作者】 任和；

【导师】 马少华；

【作者基本信息】 沈阳工业大学 ， 电机与电器， 2009， 硕士

【摘要】 电力变压器是电力传输和电网分配的核心。在很大程度上,电力变压器的运行状况直接决定电力供应的质量。因此,为确保电力供应的可靠性,有必要不断的对变压器的运行条件进行检测与评估。基于频率响应法(FRA),脉冲频率响应法(IFRA)正成为一个日益流行的用于外部检测和评估变压器绕组运行状况和结构完整性的技术。这项技术的优越性在于它可以在极短的时间内产生完整的频响特性曲线,从而使得该方法具备较强的抗干扰特性。因此,IFRA是一种理想的检测变压器绕组变形的方法。本文将以IFRA方法为基础,进行IFRA分析绕组变形方法的研究,通过OrCAD/Pspice仿真各种变压器绕组故障的状况,并使用Matlab进行各类故障状况下的频响特性分析,并绘制频响特性曲线,从而归纳出各类绕组变形频响曲线的特点。最后,进一步使用IFRA方法研制出雷电波脉冲发生器及相关辅助电路,并配以UDAQ-50612型高速采集卡搭接试验电路进行变压器绕组故障的试验研究。作为人工智能的一个分支,机器学习技术不断发展,从而使得计算机能够自主学习并完成相应的任务。本文中,通过提取频响曲线的谐振点作为特征量,使用现阶段被广泛应用的支持向量机(SVM)算法进行绕组变形的识别,从而获得变压器绕组变形程度的非线性多重分类结果。本文首次使用C语言编写出一套应用SVM算法检测绕组变形的程序集,并通过大量的仿真数据验证方法的可行性。使用Microsoft Visual Studio 2005编程开发工具的C#.NET编程语言编写试验系统的软件,包括各类试验控制模块、频响曲线生成模块、SVM试验分析模块等。最后使用本套系统对云峰发电厂1#,2#主变进行了绕组变形的识别分析。本文首次将IFRA技术与SVM算法相结合用于变压器绕组变形的分析,随着这项技术的进一步完善,有理由相信这个方法必将为变压器绕组检测做出更大的贡献。更多还原

【Abstract】 Power transformers are the core of electrical power transmission and distribution networks, and their performance determines the quality of power supply. Therefore, it’s necessary to detect winding structure in order to ensure reliability and availability of power supply.Impulse Frequency Response Analysis (IFRA), as one diagnostic method of Frequency Response Analysis (FRA) which calculates and computes frequency dependent variables of the windings, is becoming a popular technique used to externally monitor and assess the condition and mechanical integrity of transformer winding. Its superior performance is embody in its quick speed, taking only a few seconds to generate the FRA traces. Therefore, it’s an ideal diagnostic tool for detecting winding deformation even in severe conditions such as high interferences. Based on IFRA method, an analysis for transformer winding deformation is carried out. Through using OrCAD/Pspice to simulate a variety condition of transformer winding deformation and using Matlab to analyze these frequency response characteristics, the FRA traces are drawn and then the frequency response characteristics under these various fault conditions are summed up. Finally, the lighting waveform generator and the associated auxiliary circuits are developed by IFRA method, hence, the test circuit for detecting transformer winding deformation can be built with the UDAQ-50612 high-speed acquisition unit.Machine learning is considered as a subfield of Artificial Intelligence and it is concerned with the development of techniques and methods which enable the computer to learn and perform tasks and activities. In this research, the winding deformation will be identified by Support Vector Machine (SVM) algorithm, which is one of the most popular machine learning techniques and through extracting the resonant points (extreme points) of FRA traces as characteristic values to achieve the results of non-linear multi-classification. This paper uses C programming language to program a set of assembly which can detect the transformer winding deformation by SVM algorithm for the first time. And vast amounts of emulational data are used to verify the feasibility of this method. By using the C #.NET programming language of Microsoft Visual Studio 2005 development tools, the software is developed, including various types of test control modules, FRA traces generation module and SVM test analysis module.Finally, an experimental analysis for detecting transformer winding deformation is carried out to test the practicality of this detecting system through using the 1# and 2# main transformers in YunFeng power plant.It’s the first time to combine IFRA method with SVM algorithm for detecting transformer winding deformation. With the development of this technology, this method would play an important role in the filed of transformer winding deformation detecting.更多还原