基于卷积神经网络的微电网故障诊断

Fault Diagnosis of Microgrid Based on Convolutional Neural Network

微电网内部含有多种分布式电源,导致其故障时拓扑结构多变,且内部潮流双向变化,加大了微电网故障检测的难度。此前微电网故障研究多集中在网内故障方面,对网间故障提及甚少。为使微电网故障诊断更加准确、方便,提出建立一种结合卷积神经网络与微电网故障数据的故障诊断模型的方法,利用PSCAD4.6建立了微电网故障仿真系统,提取故障数据,依托Tensorflow平台进行故障诊断模型的搭建与训练。实验结果显示该故障诊断模型测试准确率高,对网内故障和网间故障,皆可对故障地点与故障类型进行检测,并具有较高的准确率。

The microgrid contains a variety of distributed power sources,which leads to a change in topology when the fault oc⁃curs,and the internal power flow changes in both directions,which increases the difficulty of microgrid fault detection.Previous studies on microgrid failures have mostly focused on intra-network failures,and little mention of inter-network failures.In order to make microgrid fault diagnosis more accurate and convenient,a method of establishing a fault diagnosis model combining convolu⁃tional neural network and microgrid fault data is proposed.Using PSCAD4.6,a microgrid fault simulation system is established to ex⁃tract fault data and rely on Tensorflow.The platform builds and trains fault diagnosis models.The experimental results show that the fault diagnosis model has a high test accuracy rate,and can detect the fault location and the fault type and the fault in the network and the network fault,and has a high accuracy rate.