With the progressive development of energy internet construction of State Grid Corpora-tion,the requirements for the reliability of power supply of distribution networks is increasingly.In order to solve the problem of low line selection accuracy of high resistance and low current ground fault in distribution networks,a high resistance and low current ground selection method based on three-dimensional space-time image is proposed.First,based on the wavelet transform,a fully convolutional neural network is used to learn the time-domain characteristics of high-resistance and low-current ground fault traveling waves,and extract the fault traveling waves.Secondly,a three-dimensional space-time image of the traveling wave time and frequency domain of a small current grounding fault is constructed,and a full convolutional neural network is used to realize the correla-tion and comparison of the traveling wave frequency of multiple faults,so as to reduce the inaccurate identification of the traveling wave frequency of a single fault.Finally,a 5 kΩ high-resistance and low-current ground fault line selection verification is carried out in a power supply company,achie-ving an accuracy rate of 95.7% .The proposed method of high resistance and low current grounding line selection based on 3D spatiotemporal image can improve the accuracy of single-phase grounding line selection and has positive significance for rapid fault disposal of distribution network.
关键词
三维时空图像/高阻/全卷积网络/小电流接地/选线方法/暂态电流行波/行波提取/透视变换
Key words
three-dimensional space-time image/high resistance/full convolutional network/low current grounding/line selection method/transient electrical popular wave/traveling wave extrac-tion/perspective transformation
维普
万方数据