神经网络法设计的差动保护重采样低通滤波器  

A Re-sampling Low-pass Filter for Differential Protection Designed with Artificial Neural Network

作　　者：陆金凤[1] 陆于平[1] 张冲[1] 

机构地区：[1]东南大学电气工程学院,江苏省南京市210096

出　　处：《电力系统自动化》2013年第8期102-106,共5页Automation of Electric Power Systems

基　　金：国家自然科学基金资助项目(50977012)~~

摘　　要：在分析了重采样方法和重采样滤波器应满足的频域特性后,提出了利用神经网络法设计有限冲激响应(FIR)重采样低通滤波器的新方法。这种方法能从频域设计具有任意幅值、相位的FIR滤波器,从而能根据前置低通滤波器的特性,对差动电流误差进行补偿,并减小这一误差,提高差动保护的可靠性。最后,建立了电力系统变压器差动保护的简化仿真模型,验证了所述理论的正确性。After analyzing re-sampling methods and frequency domain characteristic the re-sampling filter should meet,a new method for designing finite impulse response（FIR） re-sampling filter based on artificial neural network（ANN） is put forward.This method can help to design FIR filters with any amplitude or phase in frequency domain.So it can reduce the differential current error by carrying out frequency domain compensation on differential current according to the characteristics of leading low-pass filters and improve the reliability of the protection.The theory is proved by a simplified simulation.

关 键 词：差动保护 重采样 人工神经网络 频域补偿 可靠性 

分 类 号：TN713.4[电子电信—电路与系统]