风力发电系统短期功率在线预测仿真研究

摘要

In the short-term prediction of wind power,the real-time accurate prediction data are one of the most effective means to enhance the capacity of power grid to accept wind power and improve power reliability and economy.The existing prediction models based on statistical methods almost involve the preprocessing of data and training phrase,with poor adaptive ability.Considering the sparse coding method is a no-training method with high efficiency and adaptability,a class of dictionary-based sparse coding prediction method was presented.Using the dictionaries composed by the historical wind power time series data,the sparse coding prediction method or the method with elastic net regularization were applied to construct the wind power prediction model.Simultaneously,adaptive dictionary updating strategies by using three algorithms were given,so as to further improve the accuracy of the prediction.The experimental results show that different sparse coding methods have achieved good prediction results.%在风电功率短期预测研究中,实时给出精确的预测数据是增强电网接纳风电能力及改善电力可靠性与经济性的最有效手段之一.现有的基于统计方法的预测模型往往不可避免数据的预处理和模型的训练阶段,自适应能力有待提高.考虑到稀疏编码方法无需模型训练,且具有求解效率高、自适应性强的特点,提出了采用稀疏编码的短期风电在线预测模型.首先将历史风电功率时间序列数据组成具有时延的输入-输出对,分别以原子形式构成字典,再针对待预测的时延输入数据向量计算稀疏权值,进一步借用字典以得到相应的预测输出.与此同时,使用了三种自适应字典更新策略,以实现在线预测,进一步提高精度.以加拿大阿尔伯塔省的实际风电功率数据为样本,在maflab中进行了仿真.仿真结果表明,上述模型能够准确地预测短期风电功率,提高了预测的有效性和实用性.