摘要
为了解决传统电能质量分类方法提取特征困难,人工选择扰动特征和现有扰动分类准确率低、抗干扰性能差的问题,提出了一种基于胶囊网络的电能质量扰动分类方法,把扰动数据作为向量输入,通过卷积层进行特征提取、动态路由进行特征选择、数字胶囊层和全连接层进行分类,形成特征选择和分类模型相融合的分类机制。仿真结果显示,基于胶囊网络的电能质量扰动分类具有较高的准确率和良好的抗干扰能力。
In order to solve the problems of difficult feature extraction in traditional power quality classification methods,low accuracy and poor anti-interference performance of manual selection of disturbance characteristics and existing disturbance classification,a power quality disturbance classification method based on capsule network is proposed,which takes disturbance data as vector input,carries out feature extraction through convolution layer,dynamic routing for feature selection,classifies by digital capsule layer and connector layer,and a classification mechanism combining feature selection and classification model is formed.The simulation results show that the power quality disturbance classification based on capsule network has high accuracy and good anti-interference ability.
作者
刘恒
LIU Heng(College of Electronics and Information Engineering Shanghai University of Electric Power,Shanghai,201306)
出处
《红水河》
2021年第6期63-68,共6页
Hongshui River
关键词
电能质量
扰动分类
胶囊网络
深度学习
power quality
disturbance classification
capsule network
deep learning