基于趋势项分析的时间序列风速多点预测研究

MODEL BASED ON TIME SERIES PREDICTION OF WIND SPEED

基于时间序列的ARMA模型,对风速的多点预测进行研究,提出基于趋势项分析的风速预测。原始风速信号分解成趋势信号和去趋势项随机信号,并结合小波分析、滑动滤波的方法对去趋势项随机信号进行数据预处理,提高其预测精度。研究结果表明:该方法能实现满足工程需要且精度较高的风速多点预测。

The forecast of the electric energy production of a wind power station is based on the prediction accuracy of wind speed. Based on ARMA model in time series to make a multi-point prediction of wind speed, a trend-term based wind speed prediction method was proposed. In this method, the origingal wind speed signal was decomposed into trend signal and random signal, and then the no-trend random signal was pretreated with sliding filter method and wavelet analysis method to to improve the prediction accuracy. The results show that the trend-term based on wind speed prediction method can satisfy industrial requirement and achieve high-accuracy multi-point prediction of wind speed.