基于WOA-LSSVM的城市道路交通事故严重度识别

Recognition of urban road traffic accident severity based on WOA-LSSVM

为预防交通安全事故,提高城市道路交通事故严重度识别正确率和适用性,基于221起城市道路交通事故数据,选择16个城市道路交通事故严重度影响因素作为特征变量,通过对连续特征变量统计分析和离散特征变量进行赋值,构建基于WOA-LSSVM的城市道路交通事故严重度识别模型。研究结果表明:一般事故肇事者年龄、驾龄和车辆车速均值最大,分别为45岁、99个月和51.6 km/h;重大事故车辆服役时间均值最大,为57.5个月;当WOA-LSSVM模型的迭代次数为30、种群规模为300时,对应城市道路交通事故严重度识别正确率、精确率、召回率和F 1值分别为95.6%、95.3%、94.9%和94.7%,相较于LSSVM模型分别增加15.6%、16.4%、14.6%和18.3%,有效提高轻微事故识别的有效性。研究结果可为制定城市道路交通事故安全风险防控措施提供理论依据。

To prevent the traffic safety accidents and improve the correctness and applicability of severity recognition of the urban road traffic accidents,based on the data of 221 urban road traffic accidents,16 influencing factors of urban road traffic accident severity were selected as the feature variables.After the statistical analysis of continuous feature variables and assignment of discrete feature variables,a recognition model of urban road traffic accident severity based on WOA-LSSVM was constructed.The results show that the general accident perpetrators had the highest mean values of age,driving age and vehicle speed,which are 45 years,99 months and 51.6 km/h respectively.The average service time of vehicle in major accidents is the largest,which is 57.5 months.The WOA-LSSVM model has the best overall performance when the number of iterations and population size are set to 30 and 300,and the corresponding recognition correctness,accuracy,recall and F1 values of urban road traffic accident severity are 95.6%,95.3%,94.9%and 94.7%respectively,which increase by 15.6%,16.4%,14.6%and 18.3%respectively compared to the LSSVM model,so it can effectively improve the effectiveness of minor accident recognition.The research results can provide a theoretical basis for the development of safety risk prevention and control measures on the urban road traffic accidents.