短时交通流预测方法分析研究

Analysis of short-term traffic flow forecasting methods

随着国家交通路网智能化的发展,道路交通的通行管控研究已成为炙手可热的内容。高速公路交通流预测研究对缓解交通压力提高道路通行能力有着重要意义。本文通过比较LSTM模型、GRU模型和SAEs模型对都市地区高速公路交通流的预测结果,分析了三种模型的优劣及目前高速公路研究方向的不足之处,提出进一步改进措施,为交通流模型预测提供了理论分析基础。本文以真实数据为基础,分析比较了三种LSTM模型、GRU模型和SAEs模型的预测精度,研究发现SAEs模型的预测误差最小,从多个评价指标分析发现MAE降低0.14,MSE降低7.24,模型拟合度更高,预测精度显著提高,为高速公路交通流研究提供借鉴。

With the development of national traffic network intelligence,the research on traffic control of road traffic has become a hot topic.The research on expressway traffic flow prediction is of great significance to alleviate traffic pressure and improve road capacity.By comparing the prediction results of LSTM model,GRU model and SAEs model on expressway traffic flow in urban areas,this paper analyzes the advantages and disadvantages of the three models and the shortcomings of the current expressway research direction,and puts forward further improvement measures,which provides a theoretical analysis basis for traffic flow model prediction.Based on real data,this paper analyzes and compares the prediction accuracy of three LSTM models,GRU models and SAEs models.It is found that the prediction error of SAEs models is the smallest.From the analysis of multiple evaluation indicators,it is found that MAE decreases by 0.14,MSE decreases by 7.24,the model fitting is higher,and the prediction accuracy is significantly improved,which provides reference for the study of expressway traffic flow.