基于卡尔曼滤波需求预测的牛鞭效应研究

Research on Bullwhip Effect Based on Demand Forecast with Kalman Filter

需求信息失真是牛鞭效应的根源,选择合适的需求预测技术是改善牛鞭效应的关键。卡尔曼滤波因其动态更新、无需存储大量历史数据等特点,常用于非平稳需求过程的预测。基于此,构建卡尔曼滤波需求预测模型,测度不同提前期下的牛鞭效应,研究提前期及模型参数变化对牛鞭效应的影响。研究表明:提前期越长,牛鞭效应越强烈;提前期为2,多数情况下可通过卡尔曼滤波需求预测避免牛鞭效应;提前期大于2,对于同种生命周期状态的产品,无促销活动产生的牛鞭效应小于有促销活动产生的牛鞭效应,但促销力度对牛鞭效应大小无显著影响。

The distortion of demand information is the root of the bullwhip effect,and the key to improving the bullwhip effect is to select appropriate demand prediction technology.Kalman Filter is often used for the prediction of non-stationary demand processes because of its dynamic updating and no need to store a lot of historical data.Thus,the demand forecast model was obtained based on the Kalman Filter equation,and the bullwhip effect was measured at different lead time.Then the influence of the lead time and model parameters on the bullwhip effect was studied.The results showed that the longer the lead time,the stronger the bullwhip effect;In most cases,the bullwhip effect can be avoided by Kalman Filter demand forecast when lead time was 2;When the lead time was greater than 2,the bullwhip effect with no promotion activity was less than that with some promotion activities for products in the same life cycle state,but the degree of promotion had no significant effect on the bullwhip effect.