摘要
本研究旨在提出高效的继电保护方案,实现对电力系统的准确监测与保护。从状态识别与问题分类、故障检测与定位2个方面展开研究,探究深度卷积神经网络在状态识别中的应用。通过引入长短时记忆网络,实现更好的状态分类效果。文章针对过载和短路两种故障类型设计了相应的深度学习模型。通过卷积神经网络(Convolutional Neural Networks,CNN)和循环神经网络(Recurrent Neural Network,RNN)模型,实现故障位置的准确定位。实验显示,相较于传统方法,文章提出的算法在问题识别与故障定位上均取得显著改进。
This study aims to propose an efficient relay protection scheme to achieve accurate monitoring and protection of power systems.Conduct research from two aspects:state recognition and problem classification,as well as fault detection and localization,the application of deep convolutional neural networks in state recognition was explored.By introducing a long-term and short-term memory network,better state classification performance was achieved.This article designs corresponding deep learning models for overload and short circuit fault types.By using Convolutional Neural Networks(CNN)and Recurrent Neural Network(RNN)models,accurate fault location has been achieved.The experiment shows that compared to traditional methods,the algorithm proposed in the article has made significant improvements in both problem recognition and fault localization.
作者
刘泽浩
汤尔东
林忆昕
LIU Zehao;TANG Erdong;LIN Yixin(Shantou Power Supply Bureau,Shantou 515000,China)
出处
《通信电源技术》
2023年第17期68-70,共3页
Telecom Power Technology
