基于绕组电-机混合特性的变压器带电监测模型研究

It was discovered that the most faulted component of power transformer was windings, and any mechanical changes in windings would likely trigger the breakdown of the transformer. Therefore, condition monitoring of windings is needed, and it plays a very important role in preventing failures and in making the grid operate correctly. Winding is an electro-mechanical system. However, in most published approaches for condition monitoring of winding, the electrical and mechanical properties of winding were separately concerned, which may degrade the accuracy and effectiveness of the methods. This project is aimed to provide a novel online condition assessment model of winding based on its electro-mechanical properties. This model consists of two parts: online electro-mechanical information extraction part, and the health grading rules against mechanical faults of windings based on its electro-mechanical properties. In the first part, an online FRA technique will be firstly introduced to detect the changes of electric parameters by constructing the voltage-current locus diagram. Then the locus will be mapped to high-dimensional feature space to separate the electric parameters from the highly coupled equation of the locus. Besides, a method based on the Hammerstein modelling technique will be developed to extract the mechanical features of winding from the vibration - current locus diagram. In addition, experiments are designed to study the electro-mechanical properties of windings in different failure modes. Then a two-dimension feature space will be constructed to represent the health level of the winding, then classification methods will be introduced to classify the health condition of winding into several levels. The online condition assessment model of winding based on its electro-mechanical properties can be useful for monitoring the health conditions of the power transformers and windings. In this light, the online condition assessment model offer researchers and industrial practitioners a new tool to systematically capture the uncertainties of the transformer conditions, and build power in preventing failures and protecting the grid.

绕组作为电力变压器故障的最主要部件，其机械结构变化极可能会给变压器带来重大安全隐患，因此，实现运行中变压器绕组的带电监测十分必要。绕组作为一个同时具备电气特性和机械特性的特殊振动部件，目前的状态监测方法仅将其视为独立的电路系统或机械系统，从而在对绕组建立诊断模型时常出现由模型表征不完备导致的误判。本项目以研究绕组电-机混合特性及其在反映绕组状态时的作用机制，建立可应用于变压器带电监测的诊断模型为目标，开展如下研究：分别利用在线FRA和Hammerstein建模技术，研究耦合电气参数和电输入输出量的电参数方程以及绕组振动-电流关系模型，提出电、机械参数特征的在线提取方法；通过实验测量，研究电、机械参数在反映绕组故障时的规律和特点，构建故障诊断体系；根据绕组电-机特性建立可准确反映其机械结构状态的带电监测模型，为绕组振动带电监测提供了新思路，对电力变压器状态监测和电力系统安全保障具有重要意义。

绕组作为电力变压器故障的最主要部件，其机械结构的变化极可能会给变压器带来重大安全隐患，因此，实现运行中变压器绕组的带电监测十分必要。绕组作为一个同时具备电气特性和机械特性的特殊振动部件，目前的状态监测方法仅将其视为独立的电路系统或机械系统。本项目以研究绕组电-机混合特性及其在反映绕组状态时的作用机制，建立可应用于变压器带电监测的故障诊断模型为目标，开展如下研究：通过数学建模和计算机仿真，研究耦合电气参数和电输入输出量的绕组电参数模型，提出绕组电参数在线提取方法；通过实验测量，研究电参数和机械特征参数在反映绕组结构状态时的规律和关系；根据绕组电-机特性建立可准确反映绕组机械结构状态的电-振动混合模型，为绕组振动带电监测提供了新思路，对电力变压器状态监测和电力系统安全保障具有重要的理论意义和应用价值。