基于CGA-BP神经网络的光伏发电功率预测  

Prediction of Photovoltaic Power Generation Based on CGA-BP Neural Network

作　　者：朱秋璇 希望·阿不都瓦依提 ZHU Qiu-xuan;HOPE Abdulwaiti(School of Electrical Engineering,Xinjiang University,Urumqi Xinjiang 83001l,China)

机构地区：[1]新疆大学电气工程系,新疆乌鲁木齐830011

出　　处：《计算机仿真》2022年第10期113-117,共5页Computer Simulation

基　　金：新疆重大科技专项项目(2016A02004-4)。

摘　　要：针对光伏发电功率预测精度不足难以满足光伏发电并网需求的问题,提出一种基于云遗传算法(Cloud Genetic Algorithm, CGA)优化BP神经网络的短期光伏发电功率预测的方法。首先,根据天气特征选取相似日,其次,通过云模型对遗传算法的交叉概率和变异概率进行自适应调整,最后,利用CGA得到最优的权值和阈值赋值给BP神经网络的初始权值和阈值,并建立光伏发电功率预测模型。仿真结果表明,该预测模型使预测结果较其它模型效果更加理想,实现了降低预测误差的目的。Aiming at the problem that the accuracy of photovoltaic power prediction is insufficient to meet the demand of photovoltaic power grid connection, a short-term photovoltaic power prediction method based on Cloud Genetic Algorithm(CGA) optimized BP neural network is proposed. Firstly, similar days were selected according to weather characteristics. Secondly, the crossover probability and mutation probability of genetic algorithm were adaptively adjusted by cloud model. Finally, the optimal weights and thresholds obtained by CGA were assigned to the initial weights and thresholds of BP neural network, and a photovoltaic power prediction model was established. The simulation results show that the photovoltaic power predicted by CGA-BP neural network model can significantly improve the prediction accuracy.

关 键 词：光伏功率预测 相似日 云模型 云遗传算法 

分 类 号：TM743[电气工程—电力系统及自动化]