基于Matlab的时序数据两种建模和预测方法比较

TWO WAYS COMPARED ON FITTING AND PREDICTING A MODEL ABOUT TIME SERIES DATA BASED ON MATLAB

利用Matlab工具箱,对时序数据进行了两种不同形式的建模和预测。一是利用系统辨识方法来建立AR模型,利用所建模型进行预测；二是利用单隐层BP神经网络,对数据进行非线性拟合和预测。通过可视化,比较了两种方法的优劣。

Using Matlab Toolbox, two different ways were used to fit and to predict a model on time series data. One was using System- Identification- Toolbox to fit a model（ simulated and predicted）. Another was using BP- Network（ one bidden- layer） to fit a model. The advantages and disadvantages were compared under the view condition