基于PNN的变压器故障诊断模型研究

Research on Transformer Fault diagnosis based on Probabilistic Neural Networks

文中分析了神经网络在变压器故障诊断中能否切实应用并验证其是否具有独有的优势;接着阐述了建模理论及方法,建立以概率神经网络(Probabilistic Neural Networks,PNN)为理论依据的故障诊断模型,并且将该模型在Matlab环境中进行了仿真;最后还将粗糙集(Rough Set,RS)理论和PNN神经网络相结合,使得优化后的诊断模型在稳、准、快三个方面较之前得以很大的提升。

Whether neural network can be applied on transformer fault diagnosis was analyzed and its unique advantages were verified;then the theory and method of modeling were expounded,a fault diagnosis model based on PNN（Probabilistic Neural Networks）theory was established and was simulated in Matlab environment;finally,RS（Rough Set）theory and PNN neural network are used.With the combination of collaterals,the optimized diagnosis model can be greatly improved in three aspects：stability,accuracy and speed.