機車在大部分的國家都是普及率最高的交通工具,但最大的缺點在於安全性不足,機車騎士只要發生交通事故其受傷程度必較為嚴重,故為提高安全性可針對改善機車騎士的危險駕駛行為著手。 本研究於智慧型手持裝置上建立一行車資料收集平台並安裝於機車上,收集行車過程中的駕駛行為及道路資料,收集的資料透過結構化的方式儲存並透過資料倉儲技術將資料集中、清理與整理後作後續的分析。 利用資料探勘以及統計分析的相關技術找出隱藏於大料資料中的特徵值,建立一套可信賴的駕駛行為與道路資訊的辨識庫,辨識庫中除加速、減速等基本駕駛舉動外亦包含急加速、急減速、轉彎、圓環轉彎、迴轉等駕駛行為以及道路坑洞特徵值,辨識庫除可辨識駕駛行為與道路資訊外亦包含危險舉動的行車特徵門檻值。 最後,可依據辨識出來的危險舉動與道路坑洞資訊,組合成駕駛安全指標以及道路平坦度指標,作為事後督促機車騎士改善駕駛行為與道路養護的客觀依據。
In many countries, motorcycles are the most popular conveyances, the major drawback of which is lack of safety. The motorcyclists get serious injuries when occurring traffic accidents. Therefore, it’s helpful to improve the safety of the motorcyclists by improving their danger driving behaviors. This thesis focuses on constructing a data-collecting platform in the smart handheld device and assembling that device on the motorcycle to collect the data of driving behavior and road information in the driving process. The data will be structurally saved and centralized, cleaned, and systemized for further analysis through the data warehouse technology. The author uses the data-mining and statistical analysis to find the eigenvalue hidden in the mass data and to construct a reliable recognition database of the data of driving behavior and road information.In this database, besides the basic driving behaviors like speed up and speed down, it also has the other driving behaviors like the emergency acceleration, deceleration, turn, turning circle, U-turn and the eigenvalue of the potholes on the roads. The recognition database can recognize not only the data of driving behavior and road, but also the eigenvalue threshold of danger driving behaviors. Finally, by using the data of danger driving behavior and the potholes on the roads, the author will construct the the safety degree of motorcyclists and the smooth degree of roads to be the objective basis for improving motorcyclist’s driving behaviors and maintenance of the roads.