基于快速异常检出的变压器状态评估与故障诊断研究

发布时间：2018-07-07 20:43

本文选题：变压器 + 快速检出　；参考：《华北电力大学(北京)》2017年硕士论文

【摘要】：变压器的稳定运行对保障电网可靠性有重要意义,对变压器进行准确的状态评估及故障诊断是保证变压器稳定运行的重要途径。现有的变压器监测体系存在异常检出方法依靠简单的阈值法来判定是否异常,判定准确率低,存在大量误报漏报现象;变压器状态评估方法中状态量选择不够全面,状态量权重划分简单;基于油色谱的故障诊断存在检测速度较慢、准确率不足等问题。为解决上述问题,本文提出了基于快速异常检出的电力变压器状态评估与故障诊断技术的状态监测框架:先使用快速异常检出选择对设备进行状态评估和故障预测或是进行故障诊断,如果快速检出正常,则开始搜集状态参量指标情况,开始状态评估,从而辅助运维人员做出检修决策;如果快速检出异常,则对故障现象进行分析,开始进行故障诊断。首先,提出了基于油色谱时间序列的快速异常检出方法。建立了通过计算回归模型预测值与实际测量值偏差来进行快速异常检出的方法,通过对历史数据分析建立了回归模型,进而对当前值进行了预测,对比当前预测值与当前测量值偏差可以实现快速异常检出。运用实例证明了该异常检出方法的有效性。其次,基于模糊综合评价的评估方法构建了一个完整的电力变压器运行状态评估指标体系及指标评价机制,并基于综合层次分析法(The Analytic Hierarchy Process,AHP)对各个指标进行了权重确定,给出了评估等级的隶属度函数,采用了模糊集合作为评价结果,建立了变压器状态评估模型。应用建立的的模型开展了算例分析,对某市某台10k V变压器的运行状态进行了评估。再次,分析了故障类型与油中溶解气体的关系,在调研现有二分类问题方法及其核函数的基础上,选取了二叉树法及高准确率的径向基(Radical Basis Function,RBF)核函数作为模型核函数,建立了基于相关向量机(Relevance Vector Machine,RVM)的变压器故障诊断模型。应用建立的模型对样本数据进行了模拟实验,实验数据与实际案例能够很好的吻合。最终通过提出的基于快速异常检出的电力变压器状态评估与故障诊断技术状态监测框架实现了变压器的全方面监测,对保证变压器稳定运行有一定意义。
[Abstract]:The stable operation of transformers is of great significance to ensure the reliability of power networks. Accurate state evaluation and fault diagnosis of transformers is an important way to ensure the stable operation of transformers. The existing transformer monitoring system has abnormal detection methods depending on the simple threshold method to determine whether abnormal or not, the accuracy of judgment is low, there is a large number of false positives and underreporting phenomenon, the selection of state quantity is not comprehensive enough in the method of transformer condition evaluation. The fault diagnosis based on oil chromatography has some problems, such as slow detection speed and low accuracy. In order to solve the above problems, In this paper, a state monitoring framework for power transformer condition assessment and fault diagnosis based on rapid anomaly detection is proposed. If the detection is normal quickly, the condition parameters are collected and the status evaluation is started to assist the operator to make the maintenance decision. If the abnormal is detected quickly, the fault phenomenon is analyzed and fault diagnosis is started. Firstly, a rapid anomaly detection method based on oil chromatographic time series is proposed. The method of fast anomaly detection by calculating the deviation between the predicted value of regression model and the actual measured value is established. By analyzing the historical data, the regression model is established, and the current value is predicted. Rapid anomaly detection can be achieved by comparing the deviation between the current predicted value and the current measured value. The effectiveness of the method is proved by an example. Secondly, based on the fuzzy comprehensive evaluation method, a complete evaluation index system and evaluation mechanism of power transformer operation state are constructed, and the weight of each index is determined based on the Analytic hierarchy process analysis (AHP). The membership function of evaluation grade is given, and the evaluation model of transformer condition is established by using fuzzy set as the evaluation result. The operation state of a 10kV transformer in a certain city is evaluated by using the established model. Thirdly, the relationship between fault types and dissolved gases in oil is analyzed. Based on the investigation of the existing two-class problem methods and their kernel functions, the binary tree method and the radial basis function (RBF) kernel function with high accuracy are selected as model kernel functions. A transformer fault diagnosis model based on Relevance Vector Machine (RVM) is established. The model is used to simulate the sample data, and the experimental data are in good agreement with the actual cases. Finally, the state monitoring framework of power transformer based on rapid anomaly detection technology is proposed to realize all aspects of transformer monitoring, which has a certain significance to ensure the stable operation of power transformer.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM41

【参考文献】