道路交通事故死亡人数预测模型对比研究  

Comparison study on prediction models of death toll for road traffic accidents

作　　者：张嘉琦[1] 

机构地区：[1]长安大学公路学院,陕西西安710064

出　　处：《中国安全科学学报》2016年第9期45-49,共5页China Safety Science Journal

基　　金：国家自然科学基金资助(51308177;51178158);高等学校博士学科点专项科研基金资助课题(20120111120021)

摘　　要：为准确预测交通事故死亡人数,选取人、车、路和经济发展水平作为主要因素,建立GM(1,1)和Verhulst模型,进行事故预测和精度分析。结合马尔科夫方法,对已建立的模型进行修正,构建GM(1,1)-Markov,GM(1,3)-Markov和Verhulst-Markov模型。应用上述模型预测安徽省2012—2014年交通事故死亡人数。分析结果表明:应用GM(1,1)-Markov模型,3年预测值的相对误差分别为-8.4%,-12.81%和-13.18%;应用GM(1,3)-Markov模型,3年预测值的相对误差分别为-31.86%,-44.66%和-57.50%;应用Verhulst-Markov模型,3年预测值的相对误差分别为-2.68%,-2.88%和-2.42%。Verhulst-Markov模型的预测精度更高,可用来预测今后的道路交通事故死亡人数。In order to predict accurately death toll caused by road traffic accidents, the people-vehicle-road and economic development level were selected as major factors, GM （1,1） model and Verhulst model were set up, and accident prediction and accuracy analysis were carried out. combined with Markov method, the two models were corrected, GM（1,3）-Markov model, GM（1,1）-Markov model and Verhulst-Markov model were built, and road traffic accident death tolls were predicted for Anhui Province of 2012 to 2014. Analysis result shows that by using GM （1,1）-Markov model, the relative errors are - 8.4%,- 12.81%,- 13.18%,that by using GM （1,3）-Markov model, the relative errors are-31.86%,-44.66%,-57.50%,that by using Verhulst-Markov model, the relative errors are-2.68%,-2.88%,-2.42%,and that accuracy of the Verhulst-Markov model is much higher, it can be applied to the actual road traffic accident prediction.

关 键 词：交通安全 道路交通事故 灰色理论 马尔科夫理论 死亡人数 预测模型 

分 类 号：X951[环境科学与工程—安全科学]