交通事故现场安全性综合评价

为了研究交通事故现场安全性，避免二次交通事故的发生，建立了交通事故现场安全性评价指标体系，准则层为人因因素、车辆因素、道路与环境因素和管理因素4个指标，决策层共15个指标. 为了统一各评价指标的度量标准，分别采用隶属度和无量纲化函数对定性和定量指标进行无量纲处理，将各属性值变换到［0, 1］范围内. 再将处理结果作为学习样本，进行神经网络训练. 最后以典型交通事故现场有关参数的实测值输入三层BP神经网络评价模型，对交通事故现场安全性进行评价，评价结果为0.5584，安全性等级为一般. 实例应用结果表明，选取的评价指标合理，评价结果客观

Comprehensive Safety Evaluation for Traffic Accident Scene

To analyze safety of traffic accident scenes and avoid secondary accident, evaluation index system of safety in accident scene was established. Criterion layer included human factor, vehicle factor, road and environment factor, and management factor. Decision layer was consisted of 15 index. To uniform all the indexes, membership degree and dimensionless function were adopted to deal with quantitative and qualitative index. The attribute values were transformed to ［0, 1］, which trained neutral network. The BP neutral network model of three levels was developed to evaluate the safety in traffic accident scene. The result was 0.5584 and safety grade was normal. Application shows that the evaluation index is logical and the conclusion is objective.