中文摘要：

在研究已有时间序列数据流预测方法的前提下,给出了一种基于滑动窗口的时间序列数据流通用预测模型,提出能有效降噪并进行多尺度滑动窗口分析,进而进行预测的新方法Online-HHT,将数据流中的滑动窗口技术与HHT方法相结合从而达到在线分析的目的。使用此模型,通过实验证实了Online-HHT方法能够有效地对时序数据流进行在线自适应趋势预测。

英文摘要：

After studying existed methods of time series data stream,a common prediction model is given to analyze time series data stream over sliding window,a novel method named Online-HHT is presented that is with efficient de-noise ability and multi-scale analysis ability over sliding window,then prediction is maken based on the analysis results.This method connects sliding window technique in data stream field and HHT method to realize online analysis.Using this model,it approves that Online-HHT method can effectively make online self-adaption prediction for time series data stream by experiment.