基于加权移动平均的数据流预测模型*

提出一种新的基于滑动窗口的预测模型。该模型仅存储当前滑动窗口中的数据并对其进行分析，提高了计算效率；同时，为了削减在较小数据集上回归预测所产生的偏差，提出一种基于加权移动平均的数据流预测算法WMA_LRA。实验采用FDS 4.0模拟一个房屋的火灾发生情况，运用WMA_LRA算法对火灾现场的局部温度进行短期预测，结果表明该算法可以有效地提高计算效率和预测精度。

Data stream prediction model based on weighted moving average

For the sake of improving the computational efficiency,this paper proposed a new prediction model which based on sliding window in this issue. In this model,stored and processed only data in current sliding window.In order to eliminate the prediction deviation caused by small data set,proposed a stream prediction algorithm based on weighted moving average method (WMA_LRA).At last, after analyzing the house fire data stream generated by FDS 4.0,WMA_LRA algorithm shows the prediction results,and it is proved that WMA_LRA algorithm can improve the computational efficiency and the prediction accuracy efficiently.