计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (14): 249-253.DOI: 10.3778/j.issn.1002-8331.1808-0222

• 工程与应用 • 上一篇    下一篇

基于LSTM神经网络模型的交通事故预测

张志豪,杨文忠,袁婷婷,李东昊,王雪颖   

  1. 新疆大学 信息科学与工程学院,乌鲁木齐 830046
  • 出版日期:2019-07-15 发布日期:2019-07-11

Traffic Accident Prediction Based on LSTM Neural Network Model

ZHANG Zhihao, YANG Wenzhong, YUAN Tingting, LI Donghao, WANG Xueying   

  1. College of Information Science and Engineering, Xinjiang University, Urumqi 830046, China
  • Online:2019-07-15 Published:2019-07-11

摘要: 道路交通事故是道路交通安全水平的具体体现,为使预测数据更科学地为交通管理系统提供决策。提出建立基于LSTM(Long Short-Term Memory)神经网络的交通事故模型,训练交通事故相关的数据,对交通安全水平的指标进行预测。经过与传统回归模型和传统神经网络模型进行实验对比,实验显示LSTM拟合效果最佳,另外LSTM模型对同一趋势上的预测效果有明显优势。通过使用LSTM模型捕获数据中存在的时序依赖关系,能够更准确地对交通事故安全水平进行预测,使交通管理部门制定更加科学准确的决策。

关键词: 交通事故, 神经网络, 长短期记忆(LSTM), 预测, 回归

Abstract: Road traffic accident is the concrete embodiment of road traffic safety level. In order to make the prediction data more scientific, it provides decision-making for traffic management system. In this paper, a traffic accident model based on LSTM (Long Short Term Memory) neural network is proposed to train traffic accident related data and predict the traffic safety level. Compared with the traditional regression model and the traditional neural network model, the experiment shows that the LSTM model has the best fitting effect, and the LSTM model has obvious advantages in predicting the same trend. Using the LSTM model to capture the temporal dependence of the data, it can predict the safety level of traffic accidents more accurately, and make the traffic management department make more scientific and accurate decision.

Key words: traffic accident, neural network, Long Short-Term Memory(LSTM), prediction, regression