基于BNN的交通事故预警系统

A Traffic Accident Warning System Based on BNN

在国内汽车保有量逐渐增长的背景下,道路驾驶环境日趋复杂,交通事故的发生率越来越高,这对汽车驾驶员提出了更高的要求。对此,使用一种基于贝叶斯神经网络(Bayesian Neural Network,BNN)的事故预测模型预测交通事故。该模型使用行车记录仪视频进行训练,预测交通事故发生的概率。结果表明,该模型能够提前近5 s预测交通事故的发生,预测精度高达90%以上,可以实现对驾驶安全的实时监控,有助于规避驾驶过程中内部与外部的不安全因素。

With the gradual growth of domestic automobile ownership,the driving environment is becoming more and more complex,and the incidence of traffic accidents is also increasing.This has put forward higher requirements and stricter challenges to automobile drivers.In this paper,an accident prediction model based on Bayesian Neural Network(BNN)is used to predict traffic accidents.The model uses the video of dash cam for training and predicts the probability of traffic accidents.The results show that the model can predict the occurrence of traffic accidents nearly 5 s in advance,and the prediction accuracy is more than 90%.This system can achieve real-time monitoring of driving safety,which helps to avoid internal and external unsafe factors in the driving process.