摘要
传统的线性神经网络,其激励函数是固定函数,所以不能对间谐波的频率进行调整,而基波频率微小的偏差将导致谐波检测出现较大的误差。笔者在采用BP神经网络的高精度基波检测的基础之上,采用激励函数参数可调的线性神经网络来分析间谐波。理论分析和仿真结果都表明,无论有没有噪声,文中提出的方法都具有较高的检测精度,自适应能力较强,其中检测与偶数次谐波比较接近的间谐波的优势更明显。
The excitation function of traditional linear neural network is fixed,it can＇t adjust to harmonic frequency, and tiny deviation of base wave frequency will lead larger error to harmonic detection. Based on high precision of wave detection using the BP neural network, the linear neural network which has adjustable parameters activation function to analyse inter-harmonic in this paper. The simulation and the oretical analysis show that the proposed method has higher detection accuracy and adaptive ability whether with any noise, especially detecting the closed even harmonic frequency of inter-harmonic.
出处
《高压电器》
CAS
CSCD
北大核心
2013年第2期19-24,共6页
High Voltage Apparatus
基金
重庆市科技攻关计划项目(CSTC
2011AB3003)
国家"111"计划项目(B08036)~~
关键词
间谐波检测
线性神经网络
BP神经网络
inter-harmonic detection
linear neural network
back propagation neural network
