摘要
利用小波包对励磁涌流和故障电流信号进行分解并提取小波包能量特征。采用改进粒子群(PSO)算法训练概率神经网络(PNN)寻找全局最优,对PNN网络的输入输出、传递函数以及隐含层节点数进行确定,建立PNN的网络模型,对网络进行训练测试,最后提出保护判据。研究发现,该算法不仅训练速度和收敛速度快,而且具有较高的识别精度。
The algorithm applies wavelet packet to decompose excitation inrush current and fault current signals and then extract the energy characteristics of wavelet packet.Firstly,it adopts the improved PSO(particle swarm optimization)algorithm to train PNN(probabilistic neural network)to determine the input and output,the transfer function as well as the hidden layer nodes of the PNN network.Then,it establishes a network model of PNN to train and test the network.Finally,the protection criterion is proposed.The study found that not only does the algorithm have fast training speed and convergence speed,but also high recognition accuracy.
作者
公茂法
接怡冰
李美蓉
解云兴
宋健
吴娜
Gong Maofa;Jie Yibing;Li Meirong;Xie Yunxing;Song Jian;Wu Na(College of Electrical Engineering and Automation,Shandong University of Science and Technology,Qingdao 266590,Shandong,China;Zaozhuang Power Supply Company,State Grid Shandong Electric Power Company,Zaozhuang 277100,Shandong,China;Dongying Fangda Electric Power Design&Planning Company,Dongying Power Supply Company of State Grid Shandong Electric Power Company,Dongying 257091,Shandong,China)
出处
《电测与仪表》
北大核心
2018年第8期17-23,共7页
Electrical Measurement & Instrumentation
基金
国家自然科学基金资助项目(61503224)
山东省高等学校科技计划项目(J17KA074)
