传统变压器绝缘油品质荧光分析检测方法是利用荧光分光光度计采集油样本全谱段荧光光谱, 根据不同老化程度绝缘油的全谱荧光特征变化建立变压器运行状态诊断模型。 针对传统荧光方法中光度计体积大、 价格昂贵以及因光谱采集时间长无法实现实时监测等问题, 提出一种基于荧光双色比例的变压器绝缘油品质检测方法, 提取荧光特征双波段信息并建立变压器运行故障诊断模型, 可通过定制滤光片和可见光电探测器等硬件取代传统荧光分光光度计, 实现双色荧光信息快速采集及处理, 满足在线实时监测的同时降低硬件成本。 对放电击穿故障造成的变压器绝缘油老化进行荧光分析, 试验模拟了不同放电击穿工况, 制备了不同放电击穿时间(10, 30, 50, 70, 90和120 min)下的尼纳斯油样本作为荧光检测目标; 使用荧光分光光度计采集了各样本在不同激发波长下的荧光发射光谱, 确定最优固定激发波长为280 nm; 使用3点移动均值平滑法对样本荧光光谱平滑去噪, 通过分析不同放电击穿时间下油样荧光特征峰的变化规律, 选取380~388和399~407 nm作为双色信息提取波段, 利用最小二乘曲线拟合建立了荧光双色比例变压器绝缘油放电击穿故障诊断模型。 研究结果初步证明了荧光双色比例法在变压器绝缘油放电击穿故障诊断上的有效性, 为一种体积小、 成本低、 快速有效在线监测系统的建立提供了一定的理论与实践基础。

Traditional fluorescence analysis and detection methods of transformer insulated oil quality, which use fluorescence spectrophotometer to collect the full bands’ fluorescence spectrum of the oil sample and establish the diagnostic model of transformer operation states using the full bands’ fluorescence characteristics of insulating oil with a different aging degree, have the problems of large volume and the high price of photometer and the inability to realize real-time monitoring due to a long time of spectrum acquisition, a new method to detect the quality of transformer insulating oil based on the fluorescent double-color ratio to extract fluorescence characteristic dual band information and establish fault diagnosis model of transformer operation was raised to solve it. Thus, the traditional fluorescence spectrophotometer can be replaced by custom filters and visible photodetectors to realize the rapid acquisition and processing of double-color fluorescence information, which can meet the on-line monitoring and reduce the hardware cost. The aging of transformer insulating oil caused by discharge breakdown fault was analyzed by fluorescence analysis. Different discharge breakdown conditions were simulated. NYNAS oil samples with different discharge breakdown times (10, 30, 50, 70, 90 and 120 min) were prepared as fluorescence detection targets. Fluorescence emission spectra at different excitation wavelengths were collected by fluorescence spectrophotometer, and the optimal fixed excitation wavelength was found to be 280 nm. The 3-point moving mean smoothing method was used to smooth the fluorescence spectrum of the samples and by analyzing the variation of the fluorescence characteristic peak of the oil sample under different discharge breakdown times, bands of 380388 and 399407 nm were selected as the double-color information extraction band. The fault diagnosis model of transformer insulated oil discharge based on fluorescent double-color ratio was established by least-square curve fitting. The study results demonstrate the effectiveness of the fluorescent double-color method on the fault diagnosis of transformer insulated electric oil breakdown, which provides a theoretical and practical basis for establishing a small, low-cost, fast and effective online monitoring system.