基于NBC模型的电力变压器故障诊断

随着电力变压器朝着大容量、高电压、高负荷的方向发展,其结构也越发复杂,故障逐渐增加.当前广泛使用的改良三比值法在对电力变压器进行故障诊断的过程中,容易出现准确率低和漏判率高等问题.鉴于此,提出一种NBC模型,该模型将朴素贝叶斯网络和改良三比值法的优势结合,在对电力变压器当前状态数据和历史样本数据充分学习的基础上,利用朴素贝叶斯分类器对当前电力变压器故障的编码组合进行故障预测分类,将预测所得到的诊断结果进行排序整理输出,实现了对电力变压器各类故障的有效分类.仿真结果表明,该NBC模型能较好地适应电力变压器各类故障的诊断,故障综合诊断正确率为95％,相较于改良三比值法,诊断准确率提升了50％,具有较强的工程实用性.

Transformer Fault Diagnosis Based on NBC Model

With the development of power transformer towards the direction of large capacity, high voltage and high load, its structure becomes more and more complicated, and its faults increase gradually. In the process of fault diagnosis of power transformer, the improved three-ratio method, which is widely used at present, is prone to the problems of low accuracy and high miss rate.Base on the current situation, an NBC model is proposed , which combines the advantages of Naive Bayesian network and the improved three-ratio method, not only retaining the advantages of naive Na?ve Bayesian network in diagnostic structure, learning ability, classification efficiency, data processing and other aspects, but also ensuring the adaptability of the model in practical engineering. The NBC model in the current state of power transformer sample data and historical data, on the basis of fully study, using the Naive Bayesian classifier for the current of power transformer fault prediction classification code combination, will predict the diagnosis of outcomes to sort the output, implements the effective classification of all kinds of power transformer fault. The simulation results show that the NBC model has great advantages in power transformer fault diagnosis compared with the improved three-ratio method. The comprehensive accuracy of fault prediction and diagnosis is 95%, which is 50% higher than the simple improved three-ratio method in power transformer fault diagnosis. The NBC model can diagnose all kinds of power transformer faults effectively and has strong engineering practicability.