Research On Short-term Traffic Flow Prediction Technology In Terminal Area Based On Data Mining

Real-time,accurate,and efficient traffic flow prediction technology is the basis for implementing scientific air traffic flow management,and it is also the key to solving the problems of low efficiency of the air transportation system and large-scale flight delays.As the most complex subsystem,the terminal area is intensive flying,difficult to deploy,and bears huge traffic flow.With the development of the air traffic digital system,large,multisource and complex data have been generated.In this context,facing the complex terminal area,starting from the perspective of data science,which breaks the traditional thinking of relying on trajectory prediction to predict traffic flow.On the basis of the multi-source massive data such as flight operations and aviation meteorology,the temporal and spatial characteristics of the traffic flow in the terminal area have been analyzed and the data mining methods to predict the short-term traffic flow have been used.First,the Capital Airport terminal area was taken as the research object,on the basis of data processing,the regularity of the traffic flow in the time dimension was explored and analyzed,a qualitative analysis of the traffic flow influencing factors was conducted,and 7key factors that affect traffic flow were extracted through the method of correlation analysis.Secondly,based on the analysis of the terminal area traffic flow characteristics,the daily feature vector was extracted,and the similar day clustering model based on the SOM-Kmeans was constructed.And a case study was carried out to objectively summarize three typical traffic flow scenarios,which verified the effectiveness and superiority of the SOMK-means.Finally,the LSTM-BP model was constructed.And based on the classification of the daily scenes to be predicted,a short-term traffic flow prediction method combining cluster analysis and LSTM-BP was proposed.The LSTM-BP without cluster analysis,the LSTM combined with cluster analysis,and the LSTM without cluster analysis were used as comparative methods for case analysis.The results show that the method proposed is stable and has higher precision,it can achieve a more accurate prediction of the short-term traffic flow and provide decision support for controllers to carry out business work.Figure 29;Table 12;Reference 56...