论文部分内容阅读
针对现有车辆导航算法仅考虑单一数据,使所得路径实际行程时间比预期更长的问题,首先建立了基于卡尔曼滤波理论的行程时间多步预测模型;其次,提出了综合利用实时数据、行程时间多步预测数据及历史数据的实时路径导航算法,并改进了其实现的核心算法Dijkstra_pred。实验结果表明,基于三类数据的实时路径导航算法所得路径的实际行程时间从整体上优于仅采用实时数据的导航算法,且路径变化较少。
Aiming at the problem that the existing vehicle navigation algorithm only considers single data and the actual route travel time is longer than expected, a multi-step travel time prediction model based on Kalman filter theory is established firstly. Secondly, a comprehensive utilization of real-time data, Time multi-step forecast data and historical data of real-time route navigation algorithm, and improve its implementation of the core algorithm Dijkstra_pred. The experimental results show that the actual travel time of the real-time route navigation algorithm based on the three types of data is better than that of the navigation algorithm using only real-time data, and the route changes less.