摘要
风速的建模和预测对有效利用风能有着重要意义,由于风速时间序列为非正态分布且有易变性,应用统计建模的方法来精确预测风速往往较困难.本文基于一种类似于高阶马尔可夫链的Mycielski方法来预测风速,为提高预测精度,风速状态被重新定义在一个较小的范围内,然后在历史数据序列中搜寻最长长度的重复序列.数值实验和比较结果的F检验值表明改进的Mycielski方法在预测精度上得到了显著提高.
Wind speed modeling and prediction is important for utility of wind power. Since the wind speed data are non-normally distributed and have highly variable nature, it is very difficult to predict the wind speed accurately by applying statistical approaches. This paper predicts the wind speed using the Myeielski approach which is similar to a high order Markov chian approach. To improve the prediction accuracy, the wind speed states are redefined in a smaller region. Then, the Mycielski algorithm searches the longest suffix string at the end of the data sequence which had repeated at least once in the history of the sequence. The simulation examples and the F-test values of the comparsion results show that the prediction performance of the proposed approach is improved significantly.
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2013年第4期1084-1088,共5页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(61203106
60974022)
国家自然科学基金重点项目(50837001)