基于GA-WNN模型的光伏中期功率预测研究  

Mid-Term Power Prediction Study of Photovoltaic Based on GA-WNN Model

作　　者：张慧娥 刘大贵[2,3] 朱婷婷 白彩清 张慧敏 ZHANG Huie;LIU Dagui;ZHU Tingting;BAI Caiqing;ZHANG Huimin(College of Energy Engineering,Xinjiang Institute of Engineering,Urumqi 830023,China;Engineering Research Center for Enewable Energy Power Generation&Grid Technology Approved by Education Ministry(College of Electrical Engineering,Xinjiang University),Urumqi 830047,China;Power Dispatching Control Center,State Grid Xinjiang Electric Power Co.,Ltd.,Urumqi 830063,China;Inner Mongolia Extra High Voltage Power Supply Bureau,Huhehot 010080,China)

机构地区：[1]新疆工程学院能源工程学院,新疆乌鲁木齐830023 [2]可再生能源发电与并网技术教育部工程研究中心(新疆大学电气工程学院),新疆乌鲁木齐830047 [3]国网新疆电力有限公司电力调度控制中心,新疆乌鲁木齐830063 [4]内蒙古超高压供电局,内蒙古呼和浩特010080

出　　处：《自动化仪表》2024年第9期70-75,共6页Process Automation Instrumentation

基　　金：新疆维吾尔自治区自然科学基金资助项目(2020D01B18)。

摘　　要：为解决光伏发电存在限电情况下,光伏中期功率预测结果偏小导致预测精度降低的问题,提出了一种基于光伏可用功率的遗传算法(GA)优化小波神经网络(WNN)的预测模型。GA-WNN模型在预测日的相近日期内覆盖晴天、雨天、多云等多种天气类型,通过模糊C-均值聚类算法辨识限电情况,并将光伏可用功率作为训练目标,建立了WNN光伏中期预测训练模型。GA-WNN模型以预测日获取的光伏数值天气预报作为输入,经过训练后可以直接预测未来1~10 d的光伏中期功率。通过新疆某光伏运行电站的实际运行数据进行验证,预测精度达96%以上。将GA应用于WNN预测模型中,可显著提高光伏中期功率预测精度。To solve the problem of reduced prediction accuracy due to small photovoltaic medium-term power prediction results in the presence of power limitation in photovoltaic power generation,a wavelet neural network(WNN)prediction model optimized by genetic algorithm(GA)based on photovoltaic available power is proposed.GA-WNN model establishes a WNN photovoltaic medium-term prediction training model by covering a variety of weather types such as sunny,rainy,and cloudy days within the similar dates of the prediction day,recognizing the power limitation situation through the fuzzy C-mean clustering algorithm,and taking the photovoltaic available power as the training target.GA-WNN model takes the photovoltaic numerical weather forecast obtained on the forecast day as input,and after training,it can directly predict the photovoltaic medium-term power for the next 1~10 days.It is validated by the actual operation data of a photovoltaic operating power station in Xinjiang,and the prediction accuracy reaches more than 96%.The application of GA to the WNN prediction model can significantly improve the medium-term photovoltaic power prediction accuracy.

关 键 词：光伏 中期功率预测 遗传算法 小波神经网络 可用功率 模糊C-均值聚类 

分 类 号：TH-39[机械工程]