基于蚁群改进 BP算法的组合预测模型在军械器材需求预测中的应用

针对在军械器材采购计划制定环节中的器材需求测算问题,提出基于蚁群改进BP算法的组合预测模型。先结合历史数据,利用多元线性回归预测法和自回归滑动平均模型（ARMA）进行初步预测,将初步预测的结果作为蚁群改进BP网络的输入,从而得到最终的预测结果。实验结果表明,基于蚁群改进BP算法的组合预测模型能够对积累的历史数据进行充分的应用,并且有较高的预测准确性。

Application of Combined Forecasting Model in Demand Forecast of Ordnance Equipment Based on Ant Colony Improved BP Algorithm

According to equipment demand forecast in the ordnance equipment procurement planning links, a combined forecasting model of improved BP algorithm based on the colony was proposed. First combined with historical data, the multiple linear regression method and autoregressive moving average model were used to preliminary forecast, and the results of the preliminary forecast were as input of the ant colony improved BP neural network, so as to get the final results of the forecast. The experimental results showed that the im- proved BP algorithm combination forecasting model based on ant colony could be fully applied, with the accu- mulation of historical data and had higher prediction accuracy.