應用模糊分群法推估路段速率之研究

Abstract

在先進旅行者資訊系統(Advanced Traveler Information System, ATIS)中,即時路況所扮演的角色相當重要。特別是在尖峰時間或是道路壅塞時,即時路況的需求更為顯著。透過即時路況資訊的提供,可有效引導用路人選擇合適的路徑與出發間，降低運輸成本並提昇旅行效率。探針車是路況的重要來源之一,但若使用非專用探針車隊,例如公車、計程車等,因其有執行任務的需求,將會回傳與路況不一致的資料；另外,當發生嚴重擁擠而產生走走停停的現象時,所回傳的資料型態與一般狀態有著顯著的不同。本研究透過模糊分群法(Fuzzy C-Means, FCM)處理探針車回傳的資料,並在不同的路況情境下，分別建議合適的分群與計算方式。根據本研究分析結果，當路段旅行平均速率大於30km/hr(或服務水準等級為B級以上)時,可採用「分兩群取高速群」或是「分三群取高速群」的分群方式;而當路段旅行平均速率小於30km/hr(道路等級為C級以下),則建議採用「分三群取中速群」或是「直接平均」。

Providing real-time traffic information is a very important and fundamental service in ATIS. Based on this information, travelers can change their departure time or route to decrease travel cost and/or improve travel efficiency. Probe Vehicle is one of the major means to obtain traffic situation. But there are some data filtering issues need to be addressed for the non-dedicated probe vehicles, such as bus and taxi. The data generated while service is ongoing, such as passengers are getting on and off, need to be filtered out since these data do not reflect the real situation. Moreover, when the traffic is severely jammed and having the stop-and-go traffic pattern, special cares are required to rule out the unnecessary data. We proposed a Fuzzy C-Means based method to address the above mentioned issues. Our testing results show that when the average travel speed is higher than 30 km/hr, we suggest dividing the data into two or three groups and take the average of high-speed group. When the average speed is lower than 30km/hr, categorizing data into three groups and takes the middle-speed group is recommended.