基于自适应FCM和LVQ神经网络的负荷特性分类

随着电网规模日益扩大,电力负荷特性越来越多样化,精确的负荷特性分类对电力系统十分重要。基于自适应FCM和LVQ神经网络算法,提出了一种负荷特性分类方法,采用基于有效性指标函数FCM算法,产生最佳聚类数目;根据聚类结果选择最靠近每类中心的样本作为LVQ神经网络聚类的训练样本,训练学习矢量量化神经网络;通过训练好的神经网络实现对所有负荷特性样本的分类。算例分析表明是有效的和优越的。

Load Characteristic Classification Based on Self-adaptive FCM and LVQ Neural Network

With the increasing scale of power grid and more and more diversified power load characteristics,accurate classification of load characteristics becomes very important for electric power systems.Based on self-adaptive FCM and LVQ neural network algorithm, this paper presents a method for the classification of load characteristics,which uses the FCM algorithm based on the validity index function to generate the optimal number of clusters;chooses the samples closest to the center of every clustering center according to the clustering result as the training samples of the LVQ neural network to learn VQ neural network.The trained neural network is used to realize the classification of all load characteristics.The example analysis indicates that method is effective and superior.