摘要
随着新能源技术的发展和普及,大量非线性用电设备接入电网对其电能质量产生了严重影响。为解决谐波扰动信号对电力系统带来的影响,提出将改进的局部均值分解LMD(local mean decomposition)和概率神经网络相结合,构造一种电压扰动分类器,对电力系统中的电压扰动信号进行识别分类。通过构造三角波形自适应地延拓扰动信号的方法抑制LMD的端点效应,应用改进LMD算法对扰动信号进行3层分解,得到具有电压信号幅频信息的乘积函数PF(product function)分量,将由PF分量构造的信号能量作为概率神经网络的输入,以识别和分类电压干扰信号。通过建立训练模型对电压扰动信号进行仿真实验,结果表明,该方法可以准确识别电压扰动信号,有助于提高电力系统中电压扰动信号的识别精度。
With the development and popularization of new energy technologies,a large number of non-linear electrical devices are connected to grid,resulting in a serious impact on the power quality of grid.To reduce the impact of harmonic disturbance signals on power system,an approach combining improved local mean decomposition(LMD)with probabilistic neural network is proposed in this paper to construct a voltage disturbance classifier,which can identify and classify the voltage disturbance signals of power system.By formulating a triangular waveform which can adaptively extend disturbance signals to suppress the end effect in LMD,the improved LMD algorithm is used to decompose disturbance signals into three layers.In this way,the product function(PF)components with the amplitude and frequency information of voltage signals are obtained,and the signal energy value constructed by PF components is taken as input to the probabilistic neural network to identify and classify voltage disturbance signals.Simulation experiments on these signals are carried out through the establishment of a training model,and experimental results show that the proposed method can accurately identify voltage disturbance signals,which is helpful for improving the corresponding identification accuracy.
作者
牛健
张志飞
汤铭辉
赵才
王坤
NIU Jian;ZHANG Zhifei;TANG Minghui;ZHAO Cai;WANG Kun(School of Mechatronic Engineering and Automation,Foshan University,Foshan 528000,China;Guangdong Raising Synthesis Energy Services Co.,Ltd,Foshan 528000,China)
出处
《电源学报》
CSCD
北大核心
2023年第5期128-137,共10页
Journal of Power Supply
关键词
新能源
电能质量
电压扰动
局部均值分解
概率神经网络
new energy
power quality
voltage disturbance
local mean decomposition(LMD)
probabilistic neural network
