基于NEGM(1,1)与BP神经网络的广义加权比例平均备件需求组合预测

Spare Parts Demand Combination Forecast Based on NEGM(1,1)-Back Propagation Neural Networks and Generalized Weighted Functional Proportion Average

针对备件需求预测问题,构建了基于NEGM(1,1)与BP神经网络的广义加权比例平均组合预测模型,充分利用NEGM(1,1)在解决时间间隔不等的序列拟合预测问题的优越性和BP神经网络特有的计算能力强、鲁棒性好的优点,克服了单纯使用统计分析、灰色理论或神经网络的弊端.并通过实例分析,验证了此模型在备件消耗预测中的有效性与准确性.

To solve the problem of spare parts demand forecasting,the paper builds a demand combination forecast model based on NEGM （1 , 1 ）-back propagation neural networks and generalized weighted functional proportion average.It fully uses the advantages of NEGM （1,1） like solving the prediction problem of one kind of generalized non-equidistance time series and the good computing power and robustness of back propagation neural networks,and overcomes the defect of using statistic analysis,grey system or neural network solely.Finally the effectiveness and veracity of this combination model are confirmed by case studies.