组合预测模型在通信产品需求预测中的应用研究

基于通信产品多品种、小批量、需求波动的特点,本文介绍了用波形预测、移动平均、线性回归、指数平滑等4种单项预测模型拟合未来预测需求;为了克服单项预测模型的局限性、提高预测准确度,通常采用灰色关联分析法。本文提出了以关联系数和关联度的大小为依据,将4种单项预测模型加以权重系数,组合成适用于通信产品需求特点的预测模型。通过实践证明了组合预测方法的有效性,对制定主生产计划有重要指导意义。

Based on the characteristics of multi-species,small batch,and fluctuating demand of telecommunication products,this paper employs four forecasting models to fit the future forecast demand,namely wave prediction,moving average,linear regression,and exponential smoothing.The grey correlation analysis method is used to overcome the limitations of the individual model and improve the accuracy.Based on the scales of correlation coefficient and correlation degree,the four models are weighted with coefficients and combined into one forecasting model,which suits the demand of telecommunication products.Practice has shown the effectiveness of the combined forecasting model,which is of great significance to the development of the master production schedule.