摘要
针对现有风功率预测方法多为确定性的点预测,无法描述风功率的随机性的问题,建立了基于集合经验模态分解和相关向量机的短期风功率区间预测模型。首先对原始风功率序列进行集合经验模态分解,获得一个剩余分量及多个具有不同特性的固有模态分量;然后对各分量采用相关向量机算法分别建立区间预测模型;最后将各分量的预测结果进行叠加得到一定置信水平下的区间预测结果。仿真结果表明,所提的区间预测方法具有较高的预测精度和较窄区间宽度,区间覆盖率较高。
In allusion to the problem of being unable to describe randomness of wind power by present wind power prediction methods,which are mostly aiming at certain point prediction,a prediction model for short-term wind power interval based on ensemble empirical mode decomposition(EEMD) and relevance vector machine(RVM) is established.It firstly conducts EEMD on original wind power sequence to obtain a remaining component and multiple intrinsic mode functions(IMFs) with different characteristics,then respectively establishes interval prediction model for each function by using RVM algorithm,and finally overlays prediction results for getting the interval prediction result under a certain confidence level.Simulation results indicate that the proposed interval prediction method has higher prediction precision,relatively narrow interval width and higher interval coverage.
出处
《广东电力》
2016年第2期14-20,共7页
Guangdong Electric Power
基金
国家高技术研究发展计划(863计划)资助项目(2013AA050601)
关键词
短期风功率预测
区间预测
相关向量机
集合经验模态分解
混合核函数
short-term wind power prediction
interval prediction
relevance vector machine(RVM)
ensemble empirical mode decomposition(EEMD)
combined kernel function