变压器智能在线监测系统的研究

发布时间：2018-05-08 04:33

本文选题：电力变压器 + 在线监测　；参考：《哈尔滨理工大学》2017年硕士论文

【摘要】：在电网中,电力变压器数量众多且安装位置分散,存在维护成本高、不利于集中监测等问题,经常出现变压器的运行数据及运行状态无法实时监测的问题,运行过程中出现的不正常现象难以及时发现,给用户带来很大的不便,不符合国家提出的节能增效目标和国家电网公司精细化管理的要求。本文综合比较分析了目前主要变压器监测设备所侧重的监测参数及功能,在此基础上提出了一种基于多参数融合的智能变压器监测及诊断预警系统。分析了变压器智能监测预警的技术要求,根据运行监测的不同参数、安装位置及传感器类型,提出并设计了一主三从的智能在线监测系统方案;同时,对变压器的故障类型和先兆特征进行了分类研究,得到常见多发故障的先兆特征参数,根据直接获得的基本监测数据进行了能够反映故障先兆的衍生数据计算,确定了故障预警分析的特征参量,从而建立了一种新型的变压器智能监测预警系统平台,进行了相应的软硬件设计。随后,在分析比较了几种预测算法的基础上,结合变压器智能监测预警的具体特点,采用BP神经网络建立了变压器智能在线监测预警系统的数学模型,针对上述设计的系统平台采集到的变压器连续运行数据,选取了五个特征参数进行BP神经网络训练,对该组特征参数所能够反映的变压器故障进行了预测,并与实测故障先兆数据对比,仿真计算结果表明:本文提出的采用BP神经网络针对特定故障先兆特征参数的预测方法与实测故障数据基本吻合,从而证明该系统能够实现变压器智能在线故障监测及预警。通过本文所设计的智能监测预警系统,不仅能够实现变压器运行过程中几乎全部状态参数的长期连续测量,而且能够依据这些数据自动进行变压器故障先兆特征的查找、判断,并自动发出预警信号;同时还能够建立变压器运行参数大数据库,便于进一步分析变压器故障产生的原因。
[Abstract]:In the power network, the number of power transformers is numerous and the installation position is scattered, there are many problems such as high maintenance cost, which is not conducive to centralized monitoring, and the problems of transformer operation data and operation state can not be monitored in real time. The abnormal phenomenon in the operation process is difficult to find in time, which brings great inconvenience to the users, and does not accord with the goal of energy saving and increasing efficiency put forward by the state and the fine management requirements of the State Grid Company. In this paper, the monitoring parameters and functions of the main transformer monitoring equipments are compared and analyzed, and an intelligent transformer monitoring and diagnosis early warning system based on multi-parameter fusion is proposed. The technical requirements of transformer intelligent monitoring and warning are analyzed. According to the different parameters of operation monitoring, installation position and sensor type, an intelligent on-line monitoring system with one main and three followings is put forward and designed, at the same time, The fault types and precursory characteristics of transformers are classified, and the precursor characteristic parameters of common multiple faults are obtained. The derived data which can reflect the fault precursors are calculated according to the basic monitoring data obtained directly. The characteristic parameters of fault early warning analysis are determined, and a new transformer intelligent monitoring and warning system platform is established, and the corresponding software and hardware are designed. Then, based on the analysis and comparison of several prediction algorithms, combined with the specific characteristics of transformer intelligent monitoring and early warning, the mathematical model of transformer intelligent on-line monitoring and warning system is established by using BP neural network. In view of the continuous operation data of transformer collected from the system platform, five characteristic parameters are selected for BP neural network training, and the transformer faults which can be reflected by this set of characteristic parameters are predicted. Compared with the measured fault precursor data, the simulation results show that the BP neural network method proposed in this paper is in good agreement with the measured fault data. It is proved that the system can realize intelligent on-line fault monitoring and early warning of transformer. Through the intelligent monitoring and warning system designed in this paper, it can not only realize the long-term continuous measurement of almost all the state parameters in the process of transformer operation, but also automatically search and judge the precursory features of transformer faults according to these data. At the same time, the large database of transformer operation parameters can be set up, which is convenient for further analysis of the causes of transformer faults.
【学位授予单位】：哈尔滨理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM41

【参考文献】