风速时间序列的符号化描述

发布时间：2018-04-13 09:03

本文选题：风速预测 + 有条件的相关性　；参考：《电力系统自动化》2017年11期

【摘要】：利用统计分析技术,对已知的时间序列外推,可以克服短期风速预测中缺乏因果关系的困难。但在选择外推模型、参数及学习样本等方面存在主观认识模糊性的挑战。为降低主观认识模糊性对分类预测效果的影响并提高样本分类效率,提出按变化特征来定义符号,以及用符号串描述风速时间序列的粗粒化概念。在此基础上,引入趋势特征,完善风速时间序列的符号化过程,提出单元窗口特征和趋势特征相结合的两层符号化方法。利用甘肃酒泉风电基地一年的实际数据验证了该粗粒化方法的有效性。
[Abstract]:By using the statistical analysis technique to extrapolate the known time series, the difficulty of lack of causality in short-term wind speed prediction can be overcome.However, there is a challenge of subjective ambiguity in the selection of extrapolation model, parameters and learning samples.In order to reduce the influence of subjective fuzziness on classification and forecast effect and to improve the classification efficiency of samples, the concept of coarse graining of wind speed time series is proposed to define the symbols according to the variation characteristics and to describe the wind speed time series with symbol strings.On this basis, the trend feature is introduced to perfect the symbolization process of wind speed time series, and a two-layer symbolization method combining the feature of unit window and trend feature is proposed.The effectiveness of the coarse-grained method is verified by the actual data of Jiuquan wind power base in Gansu province for one year.
【作者单位】：东南大学电气工程学院;新能源与储能运行控制国家重点实验室(中国电力科学研究院);南瑞集团公司(国网电力科学研究院);智能电网保护和运行控制国家重点实验室;悉尼大学电气与信息工程学院;南方电网科学研究院有限责任公司;神华新能源责任有限公司;
【基金】：国家自然科学基金重点项目(61533010) NSFC-NRCT(中泰)合作研究项目(51561145011) 国家电网公司科技项目~~
【分类号】：TM614

【相似文献】