变压器油纸水分含量和绝缘老化程度定量评估研究

发布时间：2018-05-22 20:02

本文选题：油纸绝缘 + 特征量　；参考：《西南交通大学》2017年硕士论文

【摘要】：变压器油纸绝缘系统在热应力、机械应力及电场应力等因素作用下,其绝缘性能逐渐劣化。老化的绝缘油可以通过滤油、换新油操作使其理化性能得到改善,而绝缘纸/纸板更换难度较大,因此,变压器固体绝缘的老化状态是决定变压器剩余寿命的关键。水分既是老化产物又会加速油纸绝缘老化,如何有效区分老化和水分对频域介电谱的影响,及针对老化和水分共存的情况下,如何采取有效的方法分别评估油纸绝缘的老化程度和水分含量,是频域介电响应法更好地应用于现场变压器油纸绝缘状态评估的关键。本文在实验室内通过加速热老化试验和吸潮试验制作了不同老化程度下含水量各不相同的油浸绝缘纸试品,测试了不同状态试品的介损因数频域谱。研究表明在绝缘纸含水量较低时,可以通过影响频段的差异区分水分和老化对频域介电谱的影响。在受水分影响的特征频段内提取了介损因数对频率的积分值作为表征绝缘纸含水量的特征量,并建立了特征量与试品含水量之间的定量关系;提取了在低水分含量下有效表征油浸绝缘纸老化程度的特征量,并建立了低水分含量下特征量与试品老化程度之间的量化关系,并基于"水分补偿因子"针对高水分含量时水分对老化评估的影响进行了量化分析,最后结合"水分补偿因子"与DP及Km.c之间的拟合关系式,推导了考虑水分影响的油浸绝缘纸老化程度定量评估方法。针对老化和水分对频域介电谱的影响不易区分,频域介电谱与油浸绝缘纸老化程度、含水量之间的映射关系复杂,及提取表征水分、老化的特征量较为困难等问题,在实验条件下对不同老化程度下含水量各不相同的试品数量进行了扩充,测试了试品的介损因数频域谱与阻抗相位频域谱。通过构造多输出支持向量回归算法挖掘试品的含水量、老化程度与频域介电谱之间的复杂映射关系,然后利用其泛化能力预测油纸绝缘的水分含量和老化程度。结合自组织神经网络聚类分析结果,表明多输出支持向量回归算法在纸中含水量低于3%时,对油浸绝缘纸的含水量与老化程度能有效预测。介绍了现场变压器FDS测试的接线方式,通过X-Y模型将现场测得的变压器主绝缘结构的FDS频域谱进行换算得到绝缘纸板的FDS频谱曲线,并借助"频温平移"原理将测试温度下的FDS频谱曲线平移到参考温度下,以消除温度对测试结果的影响。建立了介损积分特征量法与多输出支持向量回归方法用于现场变压器固体绝缘状态评估的流程。最后,通过两台停运待检修的变压器FDS测试结果,验证了介损积分特征量法和多输出支持向量回归方法对现场变压器固体绝缘状态评估的有效性。
[Abstract]:Under the influence of thermal stress, mechanical stress and electric field stress, the insulation performance of oil-paper insulation system of transformer is deteriorating gradually. The physical and chemical properties of the aged insulating oil can be improved by the operation of oil filter and the replacement of new oil, but the replacement of insulator paper / paperboard is difficult. Therefore, the aging state of transformer solid insulation is the key to determine the residual life of transformer. Moisture is both a product of aging and an accelerated ageing of oil-paper insulation. How to effectively distinguish the effects of aging and moisture on dielectric spectrum in frequency domain, and aim at the coexistence of aging and moisture. How to use effective methods to evaluate the aging degree and moisture content of oil-paper insulation respectively is the key to the better application of frequency-domain dielectric response method to the evaluation of oil-paper insulation in field transformer. In this paper, through accelerated thermal aging test and hygroscopic test in laboratory, oil-immersed insulator paper samples with different water content under different aging degree were made, and the dielectric loss factor frequency spectrum of different state samples was measured. The results show that when the water content of insulator paper is low, the influence of moisture and aging on the dielectric spectrum in frequency domain can be distinguished by the difference of influence frequency band. The integral value of dielectric loss factor to frequency is extracted as the characteristic quantity of water content of insulator paper in the characteristic frequency band affected by moisture, and the quantitative relationship between characteristic quantity and water content of the sample is established. The characteristic quantity of the aging degree of oil-immersed insulator paper under low moisture content was extracted, and the quantitative relationship between the characteristic quantity and the aging degree of the sample was established under the condition of low moisture content. Based on the "moisture compensation factor", the effect of water on aging assessment is analyzed quantitatively. Finally, the fitting relationship between "water compensation factor", DP and Km.c is given. The quantitative evaluation method of aging degree of oil-immersed insulator paper considering the effect of moisture was deduced. It is difficult to distinguish the influence of aging and moisture on the frequency domain dielectric spectrum, the aging degree of frequency domain dielectric spectrum and oil impregnated insulator paper, the mapping relationship between water content, and the difficulty of extracting and characterizing moisture, and the characteristic quantity of aging, etc. Under the experimental conditions, the number of samples with different water content under different aging degree was expanded, and the dielectric loss factor frequency spectrum and impedance phase frequency domain spectrum were tested. By constructing multi-output support vector regression algorithm, the complex mapping relationship between moisture content, aging degree and dielectric spectrum in frequency domain is mined, and then its generalization ability is used to predict moisture content and aging degree of oil-paper insulation. Combined with the results of clustering analysis of self-organizing neural networks, it is shown that the multi-output support vector regression algorithm can effectively predict the moisture content and aging degree of oil-immersed insulator paper when the water content in paper is lower than 3. The connection mode of field transformer FDS test is introduced. The FDS spectrum curve of insulation paperboard is obtained by converting the FDS frequency domain spectrum of the main insulation structure of transformer measured in the field by X-Y model. Based on the principle of "frequency temperature translation", the FDS spectrum curve at the test temperature is shifted to the reference temperature to eliminate the influence of temperature on the test results. The method of dielectric loss integral eigenvalue and multi-output support vector regression are used to evaluate the solid insulation state of field transformers. Finally, through the FDS test results of two outage transformers to be overhauled, the effectiveness of the dielectric loss integral eigenvalue method and the multi-output support vector regression method for the evaluation of the solid insulation state of the field transformers is verified.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM855

【参考文献】