中图分类号：U491.1
文献标识码：A
DOI：10.19358/j.issn.2096-5133.2020.02.007
引用格式：张振，曾献辉.基于CNN-LightGBM模型的高速公路交通量预测［J］.信息技术与网络安全，2020,39（2）：34-39.

Prediction of highway traffic flow based on CNN-LightGBM model

Abstract：Effective traffic flow forecasting is of great significance to people′s travel and traffic management supervision.Traditional traffic volume prediction models are mainly based on the time characteristics of traffic flow,however,these models don′t combine the time and space characteristics of traffic flow for in-depth mining,so sometimes these models don′t perform well.This paper proposes a traffic flow prediction model based on the combination of CNN and LightGBM. The CNN model is used to excavate the temporal and spatial correlation between the monitoring points and the entrances and exits of the adjacent sections of the highway to realize the spatiotemporal feature extraction of the traffic flow data,and then the feature vector extracted by CNN is input into the LightGBM model for prediction.In order to verify the effectiveness of the model,a variety of prediction models are used in the experiment for comparison.The experimental results show that the proposed model of CNN-LightGBM considering the spatio-temporal characteristics can significantly reduce the prediction error and is an effective and fast traffic flow forecasting model.

Key words :traffic flow prediction;CNN-LightGBM; spatiotemporal correlation;highway