基于AP-WOA-GRU的分布式光伏集群电压越限动态预测  

The Dynamic Prediction of Voltage Overrun of the Distributed Photovoltaic Cluster Based on AP-WOA-GRU

作　　者：韩雨 郭成 方正云 陈凤仙 HAN Yu;GUO Cheng;FANG Zhengyun;CHEN Fengxian(School of Mechanical and Electrical Engineering,Kunming University of Science and Technology,Kunming 650500,Yunnan,China;School of Electric Power Engineering,Kunming University of Science and Technology,Kunming 650500,Yunnan,China;School of Land and Resources Engineering,Kunming University of Science and Technology,Kunming 650000,Yunnan,China;Yunnan Power Grid Co.,Ltd.,Kunming 650000,Yunnan,China;Qujing Power Supply Bureau,Yunnan Power Grid Co.,Ltd.,Qujing 655000,Yunnan,China)

机构地区：[1]昆明理工大学机电工程学院,云南昆明650500 [2]昆明理工大学电力工程学院,云南昆明650500 [3]昆明理工大学国土资源工程学院,云南昆明650000 [4]云南电网有限责任公司,云南昆明650000 [5]云南电网有限责任公司曲靖供电局,云南曲靖655000

出　　处：《电网与清洁能源》2024年第4期118-126,共9页Power System and Clean Energy

基　　金：云南省联合基金专项(202201BE070001-15);云南电网有限责任公司科技项目(YNKJXM20220074)。

摘　　要：针对整县光伏背景下规模化分布式光伏接入配电网导致的电压波动问题,提出了一种基于近邻传播聚类(affinity propagation,AP)与鲸鱼算法(whale optimization algorithm,WOA)优化门控循环单元(gated recurrent unit,GRU)的分布式光伏集群电压越限预测方法。首先,在考虑分布式光伏地理坐标气象特征的基础上,添加基于配电网节点负荷密度因素的位置特征,采用近邻传播聚类方法,在不指定聚类数目的情况下划分具有近似气象特征和地理位置特征的分布式光伏集群,提高模型训练效果及适应性;然后,采用鲸鱼优化算法全局搜索GRU模型的最优训练参数,进一步提高模型的训练速度和预测精度;最后,利用WOA-GRU组合模型实现配电网节点电压与环境温度、光照强度的关联匹配,进而实现区域配电网电压波动及电压越限情况的整体预测。实验证明:所提出的方法能够有效提高预测精度及训练速度,强化预测模型的适应能力,具有较好的经济性和实用性。Focusing on the voltage fluctuation caused by large�scale distributed photovoltaic access to the distribution network under the background of photovoltaic in the entire Qujing county,a distributed photovoltaic cluster voltage overrun prediction method based on affinity propagation(AP)and Whale Optimization Algorithm(WOA)optimized gated recurrent unit(GRU)is proposed.Firstly,the geographical coordinates meteorological characteristics and the location characteristics are added based on the node density factors of the distribution network,and the distributed photovoltaic cluster with approximate meteorological characteristics and geographical characteristics is divided by the neighbor propagation clustering method to improve the model training effect and adaptability.Secondly,the whale optimization algorithm is used to search for the optimal training parameters of the GRU model globally to improve the training speed and prediction accuracy of the model.Finally,the WOA-GRU combination model is used to realize the correlation matching between the node voltage of the distribution network with the ambient temperature and light intensity,and thus realize the overall prediction of voltage fluctuation and voltage crossing in the regional distribution network.The experiments show that the proposed method can effectively improve the prediction accuracy and training speed,strengthen the adaptability of the prediction model,and have good economy and practicability.

关 键 词：电压越限 分布式光伏 鲸鱼优化算法 门控循环单元 近邻传播聚类 

分 类 号：TM711[电气工程—电力系统及自动化]