滑动窗Walsh基函数神经网络检测电能质量扰动

摘要

为了能够准确实时检测出电力系统中存在的电能质量扰动,提出了一种自适应确定隐神经元数及修改权值的自适应Walsh基函数神经网络时频分析方法.在此基础上采用加滑动窗的方法,根据不同时刻不同基上的权值能量分布信息检测电能质量扰动的特征,实现了对电能质量扰动的实时检测.实验仿真显示该基函数神经网络有较好的逼近能力和滤噪特性,对于电能质量扰动信号的实时检测效果较好.%In order to real-time and accurately detect power quality disturbances in power system, an adaptive basic-function neural network is proposed, which can modify the number and the right of hidden basic-function adaptive. On this basis, adding a sliding window on this transformation, it can detect and recognize different power qualities disturbances by the different energy project characteristic on basic-function at different time. The simulation results show that the adaptive Walsh function neural network has good function approximation and noise filter performances and can be used to real-time detect power quality disturbances.