基于路段间转移概率的最优路径预测方法

提出一种基于路段间转移概率的最优路径预测方法，即根据出行时间，将历史出行数据分为早高峰、晚高峰、非高峰3组反映不同交通状况的时间段；根据起点和终点所在区域，对出行数据作进一步处理，解决给定起点与终点之间出行数据不足的问题。基于有经验的出行者选择路径即为最优路径这一假设，通过历史出行轨迹计算得到路段间的转移概率，建立Markov链模拟出行者路段选择行为，将最大选择概率路径作为最优路径的预测结果，并给出其求解方法。该方法仅利用历史出行轨迹进行最优路径预测，避免了复杂的路段阻抗计算，且具有数据易获取，与实际出行行为一致性高，计算量小的优点。案例分析表明，预测结果准确性在不同出行时段下产生变化，而交通区域划分的大小对其影响较小。

Predicting Optimal Route Based on Link-to-link Transition Probability

A method to find optimal route based on link-to-link transition probabilities is proposed. The trip data are divided into three groups to reflex the traffic conditions according to the departure time, namely morning peak, evening peak and off-peak; the trip data are further processed according to the zone of departure and the zone of destination, addressing the issue that there are not enough trip data for the given origin-destination. Based on the assumption that the optimal route is chosen by the experienced users, the link- to- link transition probabilities are calculated by the historical trajectories, and the Markov chain is established to simulate the link choice behavior. The route with maximum probability is suggested as the optimal route, and a calculation method is proposed. Only the historical trip data is utilized to find the optimal route by the proposal method, avoiding the complicated calculation of the link travel time. The method demonstrates the advantages of easy acquisition of data, high consistency with the actual behaviors, and low computation requirements. The case study indicates that the accuracy of proposal method may vary according to the departure time, but the sizes of traffic zones have tiny impact.