基于模糊聚类的马氏链模型在交通事故预测中的应用

Application of Markov Chain Model to the Prediction of Traffic Accidents Based on Fuzzy Clustering Analysis

运用模糊聚类分析方法,借助于MATLAB软件,对中国31个地区交通事故的危害程度进行动态分类和综合评价。根据中国1999—2005年31个地区的交通事故4个指标和5个社会经济因素数据,把31个地区分为3类,即轻灾区、较重灾区、重灾区;其中浙江、福建与宁夏的交通事故危害严重,上海、北京与天津的交通状况正在好转,这与客观实际符合。根据模糊聚类的结果,求得交通事故状态转移概率矩阵,利用马氏链模型可预测未来31个地区的交通事故发展趋势。

The severity of traffic accidents from 31 regions in China is dynamically classified and comprehensive evaluated by use of fuzzy clustering analysis and MATLAB. According to the four index and five economic data from the traffic accidents in 31 regions in China from1999 to 2005, it is divided into three categories： namely light disaster region, medium disaster region and heavy disaster region. The traffic accidents in Zhejiang Province, Fujian Province and Ningxia Province cause severe damages, and the traffics in Shanghai, Beijing and Tianjin are taking a turn for the better, which conforms to the actuality. Based on the result obtained by fuzzy clustering analysis, the probability matrix for state shift of traffic accidents was calculated. Besides, the evolutional trend of the traffic accidents in 31 regions is forecasted with a good precision through Markov chain model.