基于深度残差网络的电力系统潮流计算  

Power Flow Calculation of Power System Based on Deep Residual Network

作　　者：魏欣荣 康飞龙[1] 李佳[1] 王春光[1] 王福香[1] 魏鑫 宋志刚 Wei Xinrong;Kang Feilong;Li Jia;Wang Chunguang;Wang Fuxiang;Wei Xin;Song Zhigang(College of Mechanical and Electrical Engineering,Inner Mongolia University of Technology,Hohhot Inner Mongolia 010018,China;School of Electrical Engineering,Beijing Jiaotong University,Beijing 100044,China;Alashan Electric Power Bureau,Inner Mongolia Electric Power(Group)Co.,Ltd.,Alashan Inner Mongolia 750306,China)

机构地区：[1]内蒙古农业大学机电工程学院,内蒙古呼和浩特010018 [2]北京交通大学电气工程学院,北京100044 [3]内蒙古电力(集团)有限责任公司阿拉善电业局,内蒙古阿拉善盟750306〕

出　　处：《电气自动化》2023年第1期75-77,共3页Electrical Automation

基　　金：国家自然科学基金(32060415)。

摘　　要：为更简单、快速地进行潮流计算,提出了一种基于深度残差网络的多节点电力系统潮流算法。首先,应用仿真软件Power World Simulator中的一个典型电网实例采集样本;然后,在TensorFlow平台搭建基于深度残差网络的多节点电力系统潮流计算模型;最后,根据模型预测结果完成对方法的分析。结果表明:与传统潮流算法及基于人工神经网络的潮流算法相比,所提方法在快速性、收敛性及精度方面具有极大的优越性。In order to perform power flow calculation more simply and quickly,a multi-node power system power flow algorithm based on deep residual network was proposed.Firstly,a typical power grid example in the simulation software Power World Simulator was used to collect samples;then,a multi node power flow calculation model based on deep residual network was built on TensorFlow platform;finally,the analysis of this method was completed according to the model prediction results.The results show that compared with the traditional power flow algorithm and the power flow algorithm based on artificial neural network,this method has great advantages in terms of rapidity,convergence and accuracy.

关 键 词：电力系统 潮流计算 数据生成 TensorFlow平台 深度残差网络 

分 类 号：TM391[电气工程—电机]