摘要
为分析交通事故严重程度的影响因素,运用感知器神经网络理论,从人、车、路(环境)因素及交通流因素等方面选取12个输入参数,以交通事故严重程度为输出参数,搭建基于MATLAB平台的3层前馈人工神经网络模型。对建立的网络模型的拟合优度进行检验,求解回归曲线以及总响应,并通过实例验证模型的有效性。分析表明,驾驶员性别、年龄、事故类型对交通事故严重程度的影响可以忽略不计,交通流特征对交通事故严重程度的影响最大,天气情况、路面情况对交通事故严重程度的影响程度基本相同。
In order to analysis the influence factor of traffic accident severity, applied perceptron neural network theory, 12 input variables were selected from aspect of human, vehicle, environment and traffic flow, the output variable was the severity of accident, a three-layer feed forward artificial neural network model was built based on MATLAB platform. An empirical research was conducted by training and validation process. The goodness of fit tests were performed, the regression plots and total response were solved, the validity of model was checked through case study. Results indicate that the influence of driver＇s gender, age to the severity of accidents are negligible, the most contribution is traffic flow status, weather conditions and road conditions have equal contribution to the severity of accidents. Traffic safety and policy research must attach importance to these factors seriously.
出处
《山东交通学院学报》
CAS
2013年第2期29-34,共6页
Journal of Shandong Jiaotong University
关键词
交通事故
严重程度
神经网络
影响因素
traffic accident
severity
neural networks
influence factor
