CAJ | 학술논문

针对目前交通事故多发点鉴别常用方法存在的问题,引入了DENCLUE聚类算法用于事故多发点鉴别.对DENCLUE聚类算法的基本原理,基本定义及计算步骤进行了阐述,重点分析了该算法用于事故多发点鉴别的可行性.实例计算结果表明:与传统事故多发点鉴别法方法相比,该算法能有效的避免对排查位置进行事先划分,实现任意长度聚类;同时,在事故数据小样本的情况下,能充分凸显道路沿线的危险性,可以有效地应用于事故多发点鉴别的研究.
침대목전교통사고다발점감별상용방법존재적문제,인입료DENCLUE취류산법용우사고다발점감별.대DENCLUE취류산법적기본원리,기본정의급계산보취진행료천술,중점분석료해산법용우사고다발점감별적가행성.실례계산결과표명:여전통사고다발점감별법방법상비,해산법능유효적피면대배사위치진행사선화분,실현임의장도취류;동시,재사고수거소양본적정황하,능충분철현도로연선적위험성,가이유효지응용우사고다발점감별적연구.
@@@@Due to the existing problems associated with current accident-prone location identification methods, DENCLUE clustering algorithm was introduced to fulfill a more effective and efficient identification purpose. This paper explicated the DENCLUE algorithm about its basic rationale, primary definition, calculation process, with specially emphasizing the application to field challenges. A demonstration example in the field context was calculated with the proposed algorithm. It suggests that the algorithm can effectively avoid a pre-division of the identified locations and realize a cluster generation with random shape, comparing with those traditional methods. Furthermore, it possesses a merit to fully reflect the road risk even with small incident sample size, and thus can be applied to the research on accident-prone location identification issue.