注：黑龙江省大学生创新创业训练计划项目（No. 201910214056）
作者：刘元博，王英立，张薇
单位：哈尔滨理工大学测控技术与通信工程学院，黑龙江哈尔滨 150080
中图分类号：TK514                
文献标识码：A                
文章编号：1006-883X(2020)07-0038-05
收稿日期：2020-05-11

摘要：光伏发电输出功率受多种外界环境因素影响，存在着发电功率随机性和产能过剩的问题。针对这一问题，在分析传统BP神经网络的基础上提出了基于遗传算法优化BP（Back Propagation）神经网络的光伏预测模型。对温度、风速、压强、湿度和太阳辐射强度数据进行归纳分析，运用Matlab神经网络和遗传算法工具箱建立了基于GA-BP光伏预测算法的发电模型，并通过大量数据对算法进行训练，使得预测值与实际值误差达到最小，最后通过电路的实物搭建对该方法进行验证。该方法能提高对太阳能的利用率，同时为国家大型发电站并网问题提供参考。
关键词：光伏发电；神经网络；遗传算法；功率预测

Design of Small Solar Power Generation System Based on GA-BP Prediction Algorithm
LIU Yuan-bo, WHANG Ying-li, ZHANG Wei
Institute of Measurement and Control Technology and Communication Engineering, Harbin University of Science and Technology, Harbin 150080, China
Abstract: Due to the effects of various external environmental factors on the output power of photovoltaic power generation, there are problems of randomness of power generation and excess capacity. To solve this problem, based on a analysis of the traditional BP neural network, this paper proposes a photovoltaic prediction model to optimize BP (Back Propagation) neural network by genetic algorithm. This paper summarizes and analyzes the data of temperature, wind speed, pressure, humidity and solar radiation intensity. Using Matlab neural network and toolbox of genetic algorithm, the power generation model of GA-BP photovoltaic prediction algorithm is established, and a large number of data are used to train the algorithm to minimize the error between the predicted value and the actual value. Finally, the method is verified by circuit simulation. This method can improve the utilization rate of solar energy and provide a reference for the interconnection of large power stations in China.
Key words: photovoltaic power generation; neural network; genetic algorithm; power prediction

阅读全文

备注：2020年 第26卷 第07期