短期光伏功率模糊Elman-DIOC神经网络预测分析  

Fuzzy Elman-DIOC Neural Network Prediction and Analysis of Short-Term Photovoltaic Power

作　　者：刘媛媛 李峰[2] LIU Yuan-yuan;LI Feng(Department of Mathematics,Lvliang University,Shanxi Lvliang 033001,China;College of Electric Power Engineering,Taiyuan University of Technology,Shanxi Taiyuan 030024,China)

机构地区：[1]吕梁学院数学系,山西吕梁033001 [2]太原理工大学电力工程学院,山西太原030024

出　　处：《机械设计与制造》2024年第6期26-29,共4页Machinery Design & Manufacture

基　　金：山西省吕梁市开发区高层次科技人才引进计划专项项目(2019107);山西省研究生教育教学改革课题(2022YJJG309);山西省吕梁市科技局重点研发项目(2022GXYF14);山西省高等学校科技创新项目(2019L0983)。

摘　　要：为了提高短期光伏功率预测效率,设计了一种基于Elman神经网络的短期光伏功率预测方法。在确定网络结构与各项参数的基础上实现准确预测,该方法受到训练后表现出很高的准确率和合理性。研究结果表明:类簇数据表达到了较大相似度,类簇间数据表现出了明显差异特征。采用优化方法开展发电功率预测,建立预测值和实测值的关系,预测日达到了与实测值相近的预测结果,采用优化聚类算法获得了更精确预测结果。通过该算法优化后,预测误差均值明显下降,IMSE均值下降幅度约80%,获得更有效的聚类结果,采用优化聚类处理可以促进短期预测精度的有效提升。该研究可以拓宽到其它的同类领域中,具有很好的实际推广价值。In order to improve the efficiency of short term photovoltaic power prediction,a short term photovoltaic power predic-tion method based on Elman neural network is designed.On the basis of determining the network structure and parameters to achieve accurate prediction,the method after training shows high accuracy and rationality.The results show that:the data of class cluster express great similarity,and the data of class cluster show obvious difference characteristics.The optimized method is used to predict the power generation,and the relationship between the predicted value and the measured value is established.The predicted result is close to the measured value on the forecast day,and the optimized clustering algorithm is used to obtain more accurate prediction results.After the optimization of the algorithm in this paper,the mean value of prediction errors decreased sig-nificantly,and the mean value of IMSE decreased by about 80%,and more effective clustering results were obtained.The optimi-zation clustering processing could promote the effective improvement of short-term prediction accuracy.This research can be ex-tended to other similar fields and has good practical value.

关 键 词：光伏功率 ELMAN神经网络 预测 关联度 相似度 匹配 

分 类 号：TH13[机械工程—机械制造及自动化]