行車轉彎之行為模式辨認

The behavioral patterns of identifying with driving wheels

林煥為(Huan-Wei Lin)

指導教授 ： 姚修慎

元智大學/資訊學院/資訊工程學系/碩士(2009年)

摘要

本研究主要是利用建立一套行車轉彎辨識系統(The system of identifying with driving wheels)，以方便對於利用行車記錄器收集到的資料進行分析及辨識。由於資料量非常地大，採取方式是利用資料倉儲(Data warehouse)，將資料集中、清理與整理後作最適化調整之資訊儲存。再研究這些行車資訊所提供的資訊配合演算法做計算，找出在資訊裡面顯示的特徵值，並整理特徵資訊，提出一套可信賴的資訊辨識系統，加以應用在日常駕駛行為上。
本研究主要是針對利用行車轉彎辨識系統對轉彎行為來做探討，將其區分為行車舉動與行車特徵，行車舉動是指ㄧ般的駕駛轉彎行為，行車特徵則是去判斷ㄧ些被定義比較危險的駕駛行為，最後再用轉彎舒適度，來評比單趟車程在進行轉彎相關行為上是否會讓旅客感到不舒服。

研究結果顯示，本研究所開發出來的系統，對於ㄧ般的駕駛轉彎行為與較危險的轉彎行為有相當水準的辨識準確度，而針對轉彎舒適度的評比對於大小車都可以進行分析，長期累積下來也可以針對駕駛的行車習慣作分析，此部份可以成為業者在評斷駕駛人員優劣時ㄧ項客觀數據。

關鍵字

駕駛行為、數位式行車紀錄器、資料探勘、Pattern Rule Base

並列摘要

This research is mainly to build the system of identifying with driving wheels, and to analyze and identify the data which were collected by the traffic recorder. Since the quantity of the information is huge, so we use Data Warehouse to centralize and clear-up the information, then store the most optimal adjusted information. Following, we used algorithm to calculate the information to find the characteristics. Then use them to build a system and apply then into the driving behaviors.

This research primarily discussed the behavior of driving wheels through this system. Then separating the behavior into the “driving movement” and the “driving characteristics”. The driving movements represents the common wheeling behaviors, and the driving characteristics are defined as more dangerous wheeling behavior. Last, we use the rate of comfortable wheeling to measure that whether the single journey will make the passenger uncomfortable when processing the relative wheeling behavior.

The result of the research shows that the system that being developed by this research has the acceptable accuracy of identifying the common wheeling behavior and dangerous wheeling behavior. In measuring the rate of comfortable wheeling, we could analyze both buses and cars, and also analyze the driving habits of the drivers. This part could become an objective data that assists owners judging the drivers.

並列關鍵字

driving behavior、Digital Traffic Recorder、Data Mining、Pattern Rule Base

參考文獻

[3]游勝傑，「危險路段辨認及提醒之系統設計」，元智大學資訊工程學系，民國九十六年七月。

[7]陳垂呈，「利用有效率布林演算法探勘關聯規則」，南台科技大學資訊管理系

[11]蔡宗憲，「應用數位行車資訊結合駕駛行為與人格特質之油耗因子關聯研究」，台北大學統計學系，民國九十七年七月。

[12]楊如平，「行車資訊平台建置及資訊辨識之應用研究」，元智大學資訊工程學系，民國九十七年七月。

[15]蔡宜良，「道路特徵對行車影響之研究」，元智大學資訊工程學系，民國九十八年七月。