以數位式行車紀錄辨析高肇事大客車駕駛族群

Abstract

交通事故之發生與駕駛行為有著莫大的關係，如果能夠確實了解造成異常的原因，系統化分析駕駛行為，提早預防不當駕駛行為，進而能達到事前預防意外的發生。數位式行車紀錄器可做為提高駕駛安全一種輔助工具，透過後台統計分析軟體進行人車管理及駕駛員行車超速、急加速、急煞車等不當駕車行為之統計，判斷駕駛員的駕駛行為是否不良，對其做事前防範管理措施。本研究利用集群分析對駕駛員進行分群與管理，並透過存活分析構建駕駛事故預測模式。研究期間自民國96年1月1日至96年12月31日，以國內裝有數位式行車紀錄器大客車之某客運公司為對象，利用所有337位駕駛員的整年行車記錄，做為分析之資料來源，針對數位式行車紀錄器所記錄的駕駛行為進行統計分析；駕駛異常行為出現頻率並分別以均標、前標、頂標為門檻值進行模式建構。其中集群分析將駕駛員分為六個等級的群組，R2值達到77.1%；Cox肇事迴歸分析分別辨識出顯著影響駕駛肇事的危險因子有重度急煞車、中度急煞車、中度轉彎未減速、打左轉方向燈、冷卻水溫過高、引擎轉速過大、第七檔檔位錯誤、電磁煞車作動等八項；Cox故障迴歸分析則分別辨識出顯著影響行車故障的危險因子有冷卻水溫過高、踩離合器使車滑行、第七檔檔位錯誤等三項。

The occurrence of traffic accidents is greatly influenced by inappropriate driving behaviors. Therefore, traffic accidents may be prevented in advance if these abnormal driving behaviors could be identified, and furthermore, avoided. Digital tachograph can be utilized to enhance driving safety by providing information of driving behavior, such as over-speed, instant acceleration and braking. The driving records of 337 bus drivers between January 1, 2007 and December 31, 2007 were collected from an inter-city passenger transport service provider, which had equipped digital tachograph in its fleet. Three thresholds of the driving records in terms of frequency were used to cluster bus drivers, i.e. 50th percentile, 75th percentile and 88th percentile in normal distribution. Cluster analysis was used to classify the frequency of drivers’ abnormal driving behavior into six categories with the value of R2 reaching 0.771. The risky factors which could significantly influence the occurrence of traffic accidents were identified in Cox regression analysis, namely abrupt emergency braking, medium emergency braking, medium turning without deceleration, turning on left signal, over-heat of cooling system, over-running of engine RPM, mis-engagement of 7th gear, and electro-magnet braking. The risky factors for mechanical failures were identified as over-heat of cooling system, free-run by disengagement of clutch, and mis-engagement of 7th gear.