Application of Adaptive Kalman Filter in Logistics Demand Forecasting

The application of the Kalman filtering theory in Logistics demand forecasting is regarded as a linear system. With constant parameters and noise covariance, both the inaccuracy of forecasting models and forecasting error are reduced. Considering the variation features of Logistics demand forecasting and according to the historical data in different years, the Logistics demand model with time-variable coefficient is built up together with the observation model and system parameter model. The adaptive Kalman filtering method is applied. The empirical study shows that the adaptive Kalman filter is more accurate than both neural network algorithm and normal Kalman filter.