Power System Event Classification Based on Machine Learning
Power System Event Classification Based on Machine Learning
复制标题
基于机器学习的电力系统事件分类
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
F. Nuroğlu
中科院分区：
文献类型：
--
作者：
H. Okumus;F. Nuroğlu
为了保证电力系统的稳定和可靠运行，仅仅对发生的事件进行检测是不够的，对这些事件进行分类也是非常必要的。由于电力系统参数受到事件的高度影响，调查这些参数可以帮助分类。本文以频率扰动记录仪（FDR）记录的不同类型事件的真实的频率信号为数据集，利用机器学习对事件类型进行分类。对信号进行小波变换，然后使用不同的方法来获得特征向量，并且最好的方法给出最高的分类精度（CA）。对于事件的分类，使用随机森林（RF）、k最近邻（k-NN）和线性判别分析（LDA）。结果表明，RF分类方法是非常有效的分类事件与90.1%的CA。
To obtain a stable and reliable power system, detecting the occurring events is not enough, classifying these events are also very essential. Since the power system parameters get highly affected by the events, investigating these parameters can help in the classification. In this paper real frequency signals recorded with Frequency Disturbance Recorders (FDR) during different kind of events are used as the dataset to classify the type of the events with machine learning.For feature extraction first, the wavelet transform is applied to the signals and then different methods are used to obtain the feature vectors and the best method giving the highest classification accuracy (CA) is found. For the classification of the events Random Forest (RF), k Nearest Neighbor (k-NN) and Linear Discriminant Analysis (LDA) is used. The results show that RF classification method is highly effective in classifying the events with a 90.1 % CA.