基于负荷预测和无迹粒子滤波的配电网动态状态估计  

Dynamic State Estimation of Distribution Network Based on Load Forecasting and Unscented Particle Filter

作　　者：卢锦玲[1] 胡兴华 张学哲 王恩泽 赵增辉 LU Jinling;HU Xinghua;ZHANG Xuezhe;WANG Enze;ZHAO Zenghui(School of Electrical and Electronic Engineering,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学电气与电子工程学院,保定071003

出　　处：《电力系统及其自动化学报》2024年第4期133-140,158,共9页Proceedings of the CSU-EPSA

摘　　要：随着汽车充电成为新型重要负荷,为确保此时配电网运行与控制安全,对其进行实时准确的态势感知,提出一种基于卷积神经网络和门控循环单元的短期负荷预测与无迹粒子滤波算法自适应混合的配电网动态状态估计方法。结合使用卷积神经网络和门控循环单元进行短期负荷预测,将预测得到的有功与无功功率进行潮流计算,再与无迹粒子滤波量测估计值自适应加权得到电压幅值和相角状态估计结果。以IEEE33节点配电网为例,验证了所提状态估计方法的准确性与面对不良数据时的鲁棒性。As vehicle charging becomes a new type of important load,to ensure the safe operation and control of distribution network and conduct real-time and accurate situation awareness for the network,a dynamic state estimation method for distribution network is proposed,which is formulated by adaptively mixing the convolutional neural network(CNN)and gated recurrent unit(GRU)based short-term load forecasting and the unscented particle filter(UPF)algorithm.First,CNN-GRU is used for short-term load forecasting.Then,the predicted active and reactive power is calculated for power flow,and the result is further adaptively weighted with the actual value estimated by UPF to obtain the state estimation results of voltage amplitude and phase angle.An IEEE33 bus power distribution network is taken as an example,and the accuracy of the proposed state estimation method and its robustness in the case of bad data are verified.

关 键 词：配电网 电动汽车 负荷预测 无迹粒子滤波 动态状态估计 

分 类 号：TM74[电气工程—电力系统及自动化]