«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

27

期数:

2017年06期

页码:

130-135

栏目:

应用开发研究

出版日期:

2017-06-10

文章信息/Info

Title:

 Short-term Wind Speed Forecasting by Using Hysteretic ELM Model

文章编号:

1673-629X（2017)06-0130-06

作者:

 余敖; 陈亮; 彭敬涛

东华大学 信息科学与技术学院

Author(s):

 YU Ao; CHEN Liang; PENG Jing-tao

关键词:

 ; 极速学习机; 迟滞; 风速预测; 机器学习

Keywords:

 ; ELM; hysteresis; wind speed prediction; machine learning

分类号:

TP391

文献标志码:

A

摘要:

 为了应对能源危机,许多国家开始大力发展最有发展前景之一的风能,而风速预测是进行风电场出力预测的前提条件.目前常用的风速预测方法没有得到很高的预测精度以及预测时间.为了改善风速时间序列的预测精度和预测时间,提出了一种基于迟滞极速学习机(Extreme Learning Machine)模型的风速预测方法.ELM算法是一种新型神经网络,计算效率高,性能优越,能避免局部最小化.通过改变神经元激励函数的方式将迟滞特性引入神经网络中,以增强历史输入对当前响应的影响,从而提高有用信息的利用率,提高风速时间序列的预测性能.仿真结果表明,与ELM模型等方法相比,迟滞ELM模型能够有效减小风速时间序列的预测误差,提高了预测精度以及减少了预测时间.

Abstract:

 The wind,as one of the most promising energy,has been developed by many countries in response to the energy crisis,and the wind speed prediction is the premise of wind farm output prediction.Many wind speed prediction methods are adopted to get high prediction accuracy and prediction of time presently.In order to improve the prediction accuracy and time of wind speed time series,a wind speed prediction method based on hysteresis ELM algorithm has been proposed.ELM algorithm is a new type of neural network,and it has high computational efficiency and superior performance,which can avoid local minimum.Hysteresis characteristics have been introduced in neural network via changing neuronal excitation function to enhance influence of the response history to the current input and thus to improve the utilization rate of useful information and to promote the wind speed time series prediction performance.The simulation results show that compared with the ELM model,the hysteresis ELM model can reduce prediction error of wind speed time series effectively for improvement of prediction accuracy and decreasing of prediction time.

常用功能

更新日期/Last Update: 2017-07-28