摘要
风速时间序列具有非线性和非平稳性的特点,传统的预测方法难以建立风速间的函数关系,因此风速时间序列的预测结果精度不高。人工神经网络所具有的强非线性拟合能力有效地解决了风速时间序列难以预测的痛点,文章选择Elman神经网络预测全国3个地区不同尺度的风速时间序列,初步探讨了神经网络风速预测的可行性。结果表明,Elman神经网络经过训练,具有时序非线性拟合的能力,但预测结果精度尚未提高。
The wind speed time series has the characteristics of nonlinear and non-stationary,and the traditional prediction method is difficult to establish the functional relationship between wind speeds,so the prediction accuracy of the wind speed time series is not high.The strong nonlinear fitting ability of the artificial neural network effectively solves the pain point of the difficult prediction of the wind speed time series.This paper selects the Elman neural network to predict the wind speed time series of different scales in three regions of the country,and preliminarily discusses the feasibility of the neural network wind speed prediction.The results show that Elman neural network has the ability of time series nonlinear fitting ability after training,but the accuracy of prediction results has not been improved.
作者
李超
姜明洋
LI Chao;JIANG Mingyang(Xinjiang Goldwind Sci&Tech Co.,Ltd.,Hohhot 010010,China)
出处
《现代信息科技》
2023年第3期66-69,74,共5页
Modern Information Technology
