摘要
研究用于谐波和无功电流检测的自适应神经网络,神经网络权值根据均方误差的最快梯度下降法来调整。其本质是按误差信号与参考输入的相关性来调整,稳态时能完全检测出与参考输入相关的信号,使得被检测波形不存在波动,最后检测用数字方法实现。
Adaptive neural network for harmonic and reactive current detection is presented. The weights of neural network are adjusted according to the fastest grades of mean square error （MSE） . Weights actually are adjusted by correlation between errors and the reference inputs, which makes detection have no fluctuation when reaching steady state. Digital realizing way based on adaptive neural network is given.
出处
《电气应用》
北大核心
2006年第4期70-72,共3页
Electrotechnical Application
