汽车零部件第三方物流仓储需求量集成预测模型

A Synthetic Integrated Model to Forecast the Demand forAutomobile Parts in 3PL Warehouse

考虑到汽车零部件第三方物流企业仓储需求量的预测精度受众多非线性、不可量化的不确定因素的影响,提出一种将定量预测与定性预测相结合的集成预测模型(SIF)。SIF模型中,用RBF神经网络模型预测复杂非线性波动趋势;为补足RBF模型的若干问题,用ARIMA模型预测在库量的季节性趋势,用定性预测模型解决难以量化的外部因素的变动对需求量的影响问题。最后,将三部分结果动态叠加作为SIF模型的输出。实例分析结果表明:与各单一模型、ARIMA与RBF的组合模型相比,SIF模型具有更高的预测精度和稳定性。研究表明了SIF模型对于第三方仓储物流需求量预测的有效性和适用性。

Considering many nonlinear and unquantifiable uncertainties influencing the demand of automobile parts in3PL warehouse,an integrated forecasting model,the SIF(synthetic integrated forecasting)model,is proposed,which combines quantitative and qualitative models.In the SIF model proposed,the RBF(radial basis function)neural networks(RBF NN)predict the complex nonlinear trend of demand.For some difficulties in establishing the RBF NN,the autoregressive integrated moving average model(ARIMA)model is used to forecast the seasonal trend of demand while the experts grading method is applied to evaluate the effects of key factors that are hard to be quantified.The output of the SIF model is the dynamic integration of the predictions of these three sub-models.Case studies have shown that the SIF model has a higher accuracy compared to the sub-models and the combination of the ARIMA model and RBF NN.The research has shown the effectiveness and applicability of SIF model in forecasting the demand inventory for automobile parts in3PL warehouse service.