摘要
对比分析了事故多发点鉴别及其成因方法,归纳总结了各种方法的优缺点及适用条件。提出了基于BP神经网络的高速公路事故多发点鉴别方法,并应用实际高速公路交通事故数据,标定了网络的权值矩阵和偏置向量,该方法能有效解决矩阵法中人为确定边界的不足。针对事故多发点的突出事故诱导因素问题,提出了基于动态模糊聚类的多发点成因分析方法,该方法能将众多的事故诱导因素聚类为从主要原因到隐患原因等4个类别,进而得出较客观的成因分类结果。
The hazardous locations and the causes were compared and analyzed. The advantages, limitations, and conditions of each method were summed up. Then, a method to identify the hazardous locations based on BP neural network was established, which can overcome the limit of artificial boundary determination in matrix method. The weights and biases of the BP network were demarcated according to the data of accidents occurred on freeways. Furthermore, a method to analyze the accident causes based on dynamic fuzzy cluster was proposed, which can cluster many accident causes into four types from main causes to hidden causes. The result is objective.
出处
《交通信息与安全》
2009年第3期108-111,128,共5页
Journal of Transport Information and Safety
关键词
交通工程
事故多发点鉴别
成因分析
神经网络
动态模糊聚类
高速公路
traffic engineering
hazardous locations identification
causes analysis
neural network
dynamic fuzzy cluster
freeway
