区间离散二阶差分方程——BP神经网络组合预测方法

Combined Forecasting Method Based on Interval Discrete Second-order Difference Equation(Ⅰ-DDE) and BP Neural Network

针对小样本且具有较强波动性的区间时间序列的预测问题,文章提出了一种区间离散二阶差分方程--BP神经网络组合预测新方法,并讨论模型的相关性质,该模型对拐点区间数据具有较好的预测能力。实证预测结果表明,所提出的预测方法不但适用于小样本区间时间序列预测,对区间序列波动细节有较强的预测能力,而且比现有的区间时间序列预测模型有更高的预测精度。

For the prediction problem of interval time series with small samples and strong volatility, this paper proposes a new combined forecasting method of interval discrete second-order difference equation(Ⅰ-DDE) and BP neural network,and also discusses the related properties of the model. This model has good prediction ability for inflexion interval data. The empirical prediction results show that the proposed prediction method is not only applicable to the prediction of interval time series with small samples, but also has a strong ability to predict the fluctuation details of interval series, and has a higher prediction accuracy than other existing interval time series prediction models.