摘要
基于多层感知器可以任意精度逼近任何线性或非线性函数的基本原理 ,提出了一种采用多层前馈神经网络的频率测量方法 ,并给出相应反向传播学习算法 ( BP)神经网络的构造过程和训练方法。仿真结果表明 ,基于人工神经网络的频率测量方法具有实时性、高精度和鲁棒性 。
Because of the non linearity, randomness, distribution and non stationary of large electric power system, high measurement accuracy of frequency cannot be obtained by traditional approaches of frequency measurement. Based on Artificial Neural Network (ANN), a new approach for frequency measurement is proposed. Through a special multi layer feed forward neural network and using basic principle that multi layer network with some transfer functions can approximate any linear or nonlinear function with a finite number of discontinuities in any accuracy, a corresponding frequency measurement network is built. The building process of relevant BP neural network and the training method are given. The simulation results prove the effectiveness of the presented approach.
出处
《电网技术》
EI
CSCD
北大核心
2000年第8期40-43,共4页
Power System Technology
关键词
电网频率
测量
人工神经网络
多层感知器
Artificial Neural Network (ANN)
frequency measurement
power system
