基于二次指数平滑法和优化Kalman滤波的短时交通组合预测  

Short-term Traffic Flow Forecasting Based on Exponential Smoothing and Kalman Filter

作　　者：秦鸣[1] 杨高飞[1] 邓明君[1] 张文强[1] 冯博[1] 

机构地区：[1]华东交通大学土木建筑学院,江西南昌330013

出　　处：《北华大学学报（自然科学版）》2015年第6期814-817,共4页Journal of Beihua University（Natural Science）

基　　金：江西省自然科学基金项目(20142BAB201015);江西省科技厅科技计划项目(20123BBE50094)

摘　　要：优化改进传统的Kalman滤波模型,解决预测精度不高且预测滞后的问题.用二次平滑指数法和优化后的Kalman滤波模型两种方法组合对道路断面通过交通量进行短时交通预测,通过平均绝对误差(MAE)和平均绝对百分比误差(MAPE)分析预测结果.实例预测结果表明:当修正因子α=0.5时预测结果较为精确,同时,优化改进后的Kalman滤波模型的预测准确性有所提高.组合预测的两种方法都能够有效地降低预测误差,但按权重组合预测的效果更为明显.Optimize and improve the traditional Kalman filter model to solve the low prediction accuracy and lag. Forecast the road section by using the second exponential smoothing method and Kalman filter model method,and then analyze the forecast result with MAE and MAPE. The results show that the predicted results are relatively accurate when modified factor α= 0. 5,at the same time,the optimization of the improved Kalman filter model also can improve measurement accuracy. Combination forecast of the two methods can effectively reduce the prediction error,but the forecast effect will be more apparent by using the weight combination.

关 键 词：指数平滑法 KALMAN滤波 交通量 预测结果 预测误差 

分 类 号：U491[交通运输工程—交通运输规划与管理]