基于模糊小波神经网络的营运客车运行风险评估模型研究  

On the risk assessment model for commercial vehicles in business operation based on the fuzzy wavelet neural network

作　　者：胡立伟[1] 凌浩晗 杨锦青 赵雪亭 尹宇 田海龙 HU Li-wei;LING Hao-han;YANG Jin-qing;ZHAO Xue-ting;YIN Yu;TIAN Hai-long(Faculty of Transportation Engineering,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]昆明理工大学交通工程学院,昆明650500

出　　处：《安全与环境学报》2020年第3期862-870,共9页Journal of Safety and Environment

基　　金：国家自然科学基金项目(61863019)。

摘　　要：为提高营运客车交通安全管理水平,应准确识别营运客车在运行中的风险影响因子并对其进行评估。结合云南省2010-2017年营运客车道路交通事故数据,通过伯努利方程计算出各风险要素引发交通事故的概率并量化其造成的经济损失,在此基础上根据交通系统四要素,构建了风险评价指标体系;建立基于模糊小波神经网络的营运客车运行风险评估模型。与BP神经网络模型相比,该模型精度有所提高,收敛速度更快,学习效率更高;对模型进行实例应用,结果表明,车辆运行速度、制动器温度、胎压、不良天气、重点违法行为增长率等5个因子是造成营运客车运行高风险的重要因素,模型能够客观准确地对营运客车运行风险进行评估。The present paper is aimed at improving the traffic safety management level of the commercial vehicles through accurate identification and evaluation of the data of the road traffic accidents for the commercial business vehicles in Yunnan from 2010 to 2017. Illustrating the risk impact factors of the commercial vehicles in operation in the province,we feel it necessary to calculate the probability or likeliness in leading to the traffic accidents due to the various risk factors via the Bernoulli’s equation and then assess the economic losses caused by such accidents. And,then,efforts have to be made to calculate and work out the risk indexes of each type of such risk sources,and,as a result,it would be possible for us to deduce that the risk index of each type of such risk sources should be equal to the product of each type of the risk source leading to the accidents with the respective weighted index of the economic losses in the corresponding accidents. And,so,in accordance with the 4 elements of the traffic system,it would be possible for us to build up a risk evaluation indicator system by establishing an operational risk assessment model for the commercial vehicles based on the fuzzy wavelet neural network. Such a kind of model can also be made more accurate in comparison with the one of the BP neural network.And,what is more,the maximum absolute error of the said model can be made to be reduced by 7. 71% with the relative error being reduced by 9. 9%,in case the calculation accuracy of the model can be heightened. And,meanwhile,the length of the training time for the model has been reduced from 1 753 to 759,with the convergence speed being made faster with the training efficiency greatly elevated. And,in conclusion,the model has been made testified through a case study,so as to conclude that the above mentioned 5 important factors may lead to a great risk of the commercial vehicles,involving the vehicle speed,the brake temperature,the wheel tire pressure,the bad weather contamination and the growth rate o

关 键 词：安全管理工程 营运客车 风险识别 风险评估 模糊小波神经网络 

分 类 号：X951[环境科学与工程—安全科学]