大型变压器潜发性故障在线辨识及其柔性保护策略研究

In recent years, many power plants or power companies have a long-run winding deformation problems within 500kV transformer, and did not detect slight faults in time. They gradually develop into a serious accident causing transformer serious damage. So, it’s necessary to put forward the new requirements for abnormal operating conditions of large power transformer, slight faults detection of transformer windings and the flexible strategy of transformer outage. On the basis theory of the existing parameter identification, equipment monitoring and protection, many researches are needed to be done，such as accurate identification of various conditions of winding leakage inductance, finite element modeling of the actual transformer winding, transformer winding deformation mechanism and classification analysis, the acquisition of slight fault variation vector characteristics of excitation inductance and the smooth exit strategy for transformer after the latent faults detected, etc.The goal of this project is to enhance the online detection capability of large transformer with winding deformation and slight faults and to build a effective smooth exit strategy for transformer after the latent faults detected. The main contents include: accurate identification and validation of transformer winding leakage inductance parameters of multi state multi model, the effectiveness of finite element modeling for winding deformation, the winding deformation type classification, the vector calculation and classification research of slight faults excitation inductance, deformation characteristics, flexible transformer protection theory and test method. The results of this paper will establish a complete system for transformer latent fault detection and flexible transformer tripping in theory, which make our country at the international advanced level in this field.

近年来我国多个电厂或电力公司均出现500kV变压器绕组变形带电长期运行，以及轻微故障没有及时检出，发展为严重故障造成变压器损毁的重大事故，对大型变压器异常运行工况和绕组轻微故障的检测及变压器的柔性停运策略提出新的要求：在现有的参数辨识和继电保护理论的基础上，还须对各种工况下绕组漏感参数的精确辨识、符合实际变压器绕组的有限元建模及绕组变形机理和归类分析、轻微故障下励磁电感变化特征向量获取、检出潜发性故障后变压器的平滑退出策略等问题进行研究。本课题以全面提升大型变压器绕组变形和轻微故障的在线检测能力和潜发性故障检出后变压器平滑停运方案为着眼点，研究内容包括：变压器绕组漏感参数多状态多模型的精确辨识、绕组变形有限元建模及变形类型分类、轻微故障励磁电感变形特征向量计算及归类研究、变压器柔性保护理论及其动模方法。本课题的研究成果将从理论上建立变压器潜发性故障检测及变压器柔性跳闸的完整体系。

针对变压器潜发性故障在线监测采用的电气量和参数识别的方法准确度不够、故障特征不明显及变压器退运策略被动的问题，本项目提出了建立高精度参数辨识模型、利用绕组漏磁场故障特征判据及柔性变压器退运策略的方案，主要研究结论为：（1）建立了可用于绕组变形检测的电力变压器集总参数等效模型，用于漏感参数辨识算法的研究，并搭建了变压器绕组变形的ANSYS仿真模型，提出了依据变压器漏感参数辨识的绕组轴向变形在线检测判据；（2）通过对基于变压器最优漏磁信息的绕组变形分类方法研究，和基于漏磁场和深度信念网络的变压器绕组变形诊断方法的研究，建立了利用绕组漏磁场特征的变形异常运行工况判据；（3）通过对利用 MODWT分类原理的变压器绕组轻微故障检测研究，和利用多状态模型的变压器绕组早期故障保护方法研究，建立了基于电路模型的绕组轻微故障的检测方法；（4）建立了完全信息序贯博弈模型的大型变压器柔性保护的数学模型，通过博弈论和强化学习相结合，提出了博弈论那什均衡点的求取方法，搜索电网故障序列，并提出变压器柔性退运策略及电网防御策略,并提出了考虑变压器本体维修和电网经济性的最优停运时间确定方法；（5）并利用物理动模实验平台对本项目的参数辨识模型、潜发性故障漏磁场判据和多状态模型判据进行充分的验证；（6）在国内外权威期刊与学术会议发表学术论文20篇，其中 SCI 收录文章 1 篇，EI 收录文章 7篇；获上海市科学技术进步三等奖二项。