基于循环神经网络的电网故障智能告警信息分类研究

Research on Classification of Power Grid Fault Intelligent Warning Information Based on Cyclic Neural Network

由于电网故障告警信息较为密集,现有方法没有考虑对数据离散化处理,导致增加分类算法难度,降低对样本分类能力。提出基于循环神经网络的电网故障智能告警信息分类方法。构建混合本体集成模型,利用基于本体的方法进行电网故障智能告警信息集成,对集成到的数据信息进行去噪、填补、离散化处理,获取优化的循环神经网络结构和参数,利用循环神经网络模型实现电网故障智能告警信息分类。结果表明:所提方法得到的F1值和G-means值均要更高,分类耗时远远低于现有方法,分类平稳性较高,具有较好的实际应用价值。

Due to the dense fault alarm information,the existing methods do not consider the data discretization,which results in the increase of classification algorithm difficulty and the reduction of sample classification ability.This paper presents a method of fault intelligent alarm information classification based on recurrent neural network.It builds a hybrid ontology integration model,use the ontology based method to integrate the intelligent alarm information of power grid fault,denoise,fill and discretize the integrated data information,obtains the optimized structure and parameters of the cyclic neural network,and uses the cyclic neural network model to realize the intelligent alarm information classification of power grid fault.The results show that the F1 value and g-means value of the proposed method are higher,the classification time is far lower than that of the existing method,the classification stability is higher,and it has better practical value.