摘要
为兼顾操作冲击电压、雷电冲击电压作用下对CVT的补偿效果,针对CVT的操作、雷电冲击响应所含的频率分量相差较大,本文提出了一种基于分类器和双路BP神经网络还原CVT一次侧波形的方法。通过设置频率分点计算高于频率分点的小波能量曲线以区分冲击电压类型,进而将冲击电压响应输入对应的双路BP神经网络神经元。通过多组雷电、操作冲击电压波形对神经网络进行训练,并验证了该方法能够兼顾CVT的非线性,模型具有波形适应性。现场应用表明,该方法可以用来对CVT进行暂态电压传递特性补偿,为暂态电压监测与电力故障分析提供理论指导。
For taking into account the compensation effect of CVT under the action of switching impulse voltage and lightning impulse voltage and in view of such issue as large frequency component difference,which is contained in the lightning impulse response in the CVT operation,a kind of method based on a classifier and a two-way BP neural network to revert the CVT primary waveform is proposed in this paper. The wavelet energy curve higher than the frequency point is calculated by setting the frequency point to distinguish the type of impulse voltage,and then the impulse voltage response is input to the corresponding two-way BP neural network neuron. The neural network is trained through multiple sets of lightning and switching impulse voltage waveforms. It is verified that the method can take into account the non-linearity of CVT and the model has waveform adaptability. The field application shows that the method can be used to compensate the transient voltage transfer characteristics of CVT and provide theoretical guidance for transient voltage monitoring and power failure analysis.
作者
谢施君
雷汉坤
王乃会
潘曦宇
刘毅
林福昌
XIE Shijun;LEI Hankun;WANG Naihui;PAN Xiyu;LIU Yi;LIN Fuchang(State Grid Sichuan Electric Power Research Institute,Chengdu 610014,China;State Grid Panzhihua Power Supply Company,Sichuan Panzhihua 617000,China;State Grid Leshan Power Supply Company,Sichuan Leshan 614000,China;SEEE,AEET,Huazhong University of Science&Technology,Wuhan 430074,China)
出处
《高压电器》
CAS
CSCD
北大核心
2022年第2期149-157,共9页
High Voltage Apparatus
基金
国网四川省电力公司科技项目(52199717001Y)。
