计算机科学与探索

• 学术研究 •    

风电输出功率预测技术研究综述

武煜昊, 王永生, 徐昊, 陈振, 张哲, 关世杰   

  1. 1.内蒙古工业大学 数据科学与应用学院, 呼和浩特 010080
    2.内蒙古自治区基于大数据的软件服务工程技术研究中心, 呼和浩特 010080
  • 出版日期:2022-08-12 发布日期:2022-08-12

Survey of Wind Power Output Power Forecasting Technology

WU Yuhao, WANG Yongsheng, XU Hao, CHEN Zhen, ZHANG Zhe, GUAN Shijie   

  1. 1.College of Data Science and Application, Inner Mongolia University of Technology, Hohhot 010080, China
    2.Inner Mongolia Autonomous Region Engineering & Technology Research Center of Big Data Based Software Service, Hohhot 010080, China
  • Online:2022-08-12 Published:2022-08-12

摘要: 风电具有的波动性、间歇性等特点对并网造成一定程度的影响,提前进行风电功率预测是解决上述问题的一个重要途径。但传感器传输、网络通信等不可控因素的存在,导致采集到用于风电功率预测的数据存在异常值和缺失值,因此在进行风电功率预测前应当进行相应的异常值检测和缺失值插补操作。为进一步促进风电数据清洗及预测技术的发展,对当前现有模型及方法进行分析与总结,并对现有技术进行划分、对比。从时序数据出发,首先对风电预测领域的异常值检测方法的研究现状进行分类、分析与总结,对现有异常检测方法所存不足与缺陷进行概述,并对未来发展中或将成为重点的研究方向进行展望;其次,将现有的缺失值处理方法的评价指标进行描述,根据处理方式的不同将处理技术按照常规处理方法、辨别式的插补方法、生成式的插补方法及物理特性方法进行分析与总结,并对现有研究中所存问题进行分析;最后,对现有研究中的预测方法、多层级预测及自适应预测系统的研究现状进行分析总结,并对现有预测存在的挑战及未来发展方向进行了总结与展望。

关键词: 深度学习, 风电功率预测, 异常值检测, 缺失值插补, 时间序列数据

Abstract: However, due to the uncertainty and volatility of wind power generation, make the gird-connected wind power system present some serious challenges. Prediction wind power in advance is an important way to solve the above problems. Due to the existence of uncontrollable factors such as sensor transmission and network communication, the data collected for wind power prediction have abnormal values and missing values. Therefore, corresponding outlier detection and missing value interpolation operations should be performed before wind power prediction. Starting from time series data, this paper first classifieds, analyzes and summarizes the research status of outlier detection methods in the field of wind power prediction, summarizes the deficiencies and defects of existing anomaly detection methods, and prospects the research directions that may become the focus in the future development. Secondly, the evaluation indexes of the existing missing value treatment methods are described. According to the different treatment methods, the processing techniques are analyzed and summarized according to the conventional treatment methods, discriminative interpolation methods, generative interpolation methods and physical characteristics methods, and the existing problems in the existing research are analyzed. Finally, the current research status of forecasting methods, multi-level forecasting and adaptive forecasting systems in existing research are analyzed and summarized, and the existing challenges and future development directions of existing forecasting are summarized and prospected.

Key words: deep learning, wind power forecasting, outlier detection, missing value imputation, time series data