干式变压器局部放电在线监测及模式识别的研究

发布时间：2019-05-28 05:34

【摘要】：干式变压器作为电力变压器中的一种,常应用于特殊的场合,其重要性不言而喻。局部放电不仅是造成干式变压器绝缘劣化和破坏的主要因素,也是反映其绝缘缺陷的重要征兆;局部放电存在不同的类型,不同的放电类型是由不同的绝缘缺陷引起的,如果能判断出局部放电的类型可以有助于更加迅速确定变压器的绝缘缺陷性质和部位,方便检修工作的展开。因此对干式变压器的局部放电在线监测和模式识别技术进行研究具有很重要的意义。首先,本文对局部放电发生的机理和局部放电发生时超声波产生的原因进行了相关分析,以及对干式变压器的绝缘缺陷类型进行研究,发现造成其绝缘缺陷的因素主要有原材料材质的选择、绝缘结构的设计和绕组浇注成型工艺等,对应会导致针尖放电、沿面放电和气隙放电等三种放电的发生;针对这三种放电类型,制作了三种放电模型,并在实验室条件下在干式变压器内部模拟了三种放电的发生,得到了局放超声信号的原始数据。其次,由于小波包变换理论非常适合非平稳信号的分析,因此本文对小波包分析理论进行相关研究,利用小波包分析理论对局部放电超声信号进行分析,提取出包含局部放电绝大部分信息能量的小波包分解系数,用这些系数的统计量作为模式识别的特征输入向量,最后利用概率神经网络实现了三种放电类型的模式识别。最后,在虚拟仪器平台上开发了基于超声波信号的干式变压器局部放电在线监测与故障预警系统,该系统是以LabVIEW图形化编程语言为主,并结合Matlab Script节点中的Matlab工具,实现了对局部放电超声信号的数据采集保存、放电波形的实时显示、放电阈值设置及故障的预警等,另外还提供对保存数据的分析功能,主要可实现局部放电数据的历史波形再现、局放趋势分析、局放源的定位及局放类型的模式识别等功能,为保障干式变压器的安全运行向运行人员提供了丰富的局部放电信息,方便了缺陷的确定和检修。该系统现已成功应用于海上钻井平台主变压器上,运行效果良好。
[Abstract]:As a kind of power transformer, dry transformer is often used in special occasions, and its importance is self-evident. Partial discharge is not only the main factor causing insulation deterioration and destruction of dry transformer, but also an important sign to reflect its insulation defects. There are different types of partial discharge, and different discharge types are caused by different insulation defects. If we can determine the type of partial discharge, it can help to determine the nature and location of insulation defects of transformer more quickly. Facilitate the development of maintenance work. Therefore, it is of great significance to study the on-line partial discharge monitoring and pattern recognition technology of dry transformer. First of all, this paper analyzes the mechanism of partial discharge and the causes of ultrasonic wave when partial discharge occurs, and studies the types of insulation defects of dry transformer. It is found that the main factors causing the insulation defects are the selection of raw materials, the design of insulation structure and the casting process of windings, which will lead to needle tip discharge, surface discharge and air gap discharge. According to these three discharge types, three discharge models are made, and the occurrence of three kinds of discharge is simulated in the dry transformer under laboratory conditions, and the original data of partial discharge ultrasonic signal are obtained. Secondly, because the wavelet packet transform theory is very suitable for the analysis of non-stationary signals, this paper studies the wavelet packet analysis theory, and uses the wavelet packet analysis theory to analyze the partial discharge ultrasonic signal. The wavelet packet decomposition coefficients containing most of the information energy of partial discharge are extracted, and the statistics of these coefficients are used as the feature input vectors of pattern recognition. finally, the probabilistic neural network is used to realize the pattern recognition of three discharge types. Finally, an on-line partial discharge monitoring and fault warning system for dry transformer based on ultrasonic signal is developed on the virtual instrument platform. The system is based on LabVIEW graphical programming language and combined with Matlab tool in Matlab Script node. The data acquisition and preservation of partial discharge ultrasonic signal, the real-time display of discharge waveform, the setting of discharge threshold and the early warning of fault are realized, and the analysis function of saved data is also provided. It can mainly realize the historical waveform reproduction of partial discharge data, the analysis of local discharge trend, the location of partial discharge source and the pattern recognition of local discharge type, which provides abundant partial discharge information to operators to ensure the safe operation of dry transformer. It is convenient to determine and repair the defects. The system has been successfully applied to the main transformer of offshore drilling platform, and the operation effect is good.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM412

【相似文献】