In this paper, a bus travel time prediction model is present which based on GPS data. In this model, weight dynamic distribution is used to balance the real data and historical data. ARIMA model is used to value the influence factor of previous buses which just past on the current bus. It can effectively eliminate the disturbance of the bus bunching. This model is proved to be accurate and stable in real test on bus line 105 in Beijing. The average accuracy is up to 88. 4%.%基于公交车辆的GPS定位信息,利用公交站点间行程时间的实时数据与短期历史数据相结合的方法,设计权重动态分配的公交到站时间预测模型,利用ARIMA模型动态确定前几辆车的行程时间对当前车辆的不同影响权重,来预测当前车辆的行程时间,能够有效的消除公交串行事件对预测的干扰.该预测模型通过北京市某路公交线路实时数据接入测试验证平均准确度达到88.4％,表明该预测模型有较高的准确性与可靠性.
展开▼
