基于改进相似性搜索算法的短期风速预测  

Short-Term Wind Speed Forecasting Based on Improved Similarity Search Algorithm

作　　者：王敏[1] 潘永春[1] 李想[1] 陈芬[1] 傅质馨[1] 

机构地区：[1]河海大学能源与电气学院,江苏南京211100

出　　处：《陕西电力》2016年第3期1-5,共5页Shanxi Electric Power

基　　金：国家自然科学基金--青年科学基金项目资助(51307044);教育部中央高校基本科研业务费专项资金资助项目(2012B05914)

摘　　要：风速及其波动特性的预测对包含风电场的电力系统运行有着重要意义,可以有效地减轻或避免风电对电力系统的不利影响,从而提高风电在电网中的渗透功率,提高电力系统的风电消纳能力。依据对风电波动性和变化趋势性进行的分析,提出采用改进相似性搜索法选择待预测数据的时序神经网络的训练样本,基于我国沿海某风场的风速预测算例表明该方法有效地提高了模型的预测精度,扩大了模型的预测范围,强化了模型的适应性。Wind speed forecasting and its variability characteristics are very important to the operation of the power systems with wind power plants.It can relieve or avoid the disadvantageous impact of wind power plants on power systems,improve the wind energy penetration in the grid,and enhance the competitive ability of wind power.Improved similarity search algorithm is chosen as the neural network training samples.The algorithm is improved based on the analysis of variability characteristics and variation tendency of wind series.The wind prediction result shows that the forecasting accuracy of model increases and the generalization ability of model enhances in a inshore wind farm in China.

关 键 词：短期风速预测 神经网络 相似性搜索法 波动特性 

分 类 号：TM614[电气工程—电力系统及自动化]