信息粒化和相似数据结合的短期风速预测  

Short-term Wind Speed Forecasting Based on Information Granulation and Similar Data

作　　者：廖玲珑 李培强[1] 唐捷[2] 李欣然[1] 李慧[1] 李文英[1] 

机构地区：[1]湖南大学电气与信息工程学院,长沙410082 [2]广东电网公司韶关供电局,韶关512000

出　　处：《电力系统及其自动化学报》2016年第4期12-17,共6页Proceedings of the CSU-EPSA

基　　金：国家自然科学基金资助项目(51277055)

摘　　要：准确地短期风速预测对电网安全稳定具有重要意义。该文将信息粒化和相似数据搜索应用于风速预测中,实现对风速变化区间的预测。对原始风速数据用模糊信息粒化处理,在此基础上运用新的相似数据搜索法和支持向量机结合的模型进行预测,得到变化区间。实例仿真表明,粒化后的数据反映了风速变化特征,减少了数据的冗余,新的相似数据搜索法提高预测模型训练样本和预测样本的相关性,提高了预测精度。通过风电场实际风速数据验证,该模型能有效预测风速的变化范围。Accurate short-term wind forecasts for the grid security and stability is of great significance. In this paper, information granulation and similar data search are applied to predict the range of wind speed. Wind data is processed with the theory of fuzzy information granulation. On this basis support vector machine （SVM） with similar data search is used to forecast. Finally, the range is obtained. The simulation results show that granulated data can not only reflect the characteristics of wind but also reduce redundant information. The new similar data search method improves the correlation of the forecasting samples and the training samples, and the prediction accuracy is improved. By validation with the actual wind speed data of a wind farm, this model can effectively predict the range of wind speed.

关 键 词：风速变化区间 信息粒化 相似数据搜索 支持向量机 

分 类 号：TK81[动力工程及工程热物理—流体机械及工程]