Airport Surface Traffic Situation Prediction Based On Spatio-Temporal Characteristics Method Research

The rapid growth of China’s civil aviation industry had led to an increase in traffic demand at some of the major airports that exceeded the capacity of the airport system.This had resulted in traffic congestion shifting from the airspace to the airport scene,causing flight delays and reducing airport operation efficiency.To address this issue,it was crucial to predict the traffic situation on the airport surface in advance,while taking into account the limitations of existing airport resources and capacity.This prediction could serve as a scheduling basis for the airport and enhance the operation efficiency of the airport surface.In this context,the focus of this paper was on predicting the airport surface traffic situation.The study considered the space-time characteristics of the airport surface,simulated the space-time impact factors that affected the airport surface traffic situation,and built a combined prediction model of convolutional neural network(CNN)and long short-term memory(LSTM)to predict and classify the airport surface traffic situation.The primary objective of this research was to develop a model that accurately predicted the airport surface traffic situation,thus improving the operational efficiency of the airport.The main work of this paper was as followed:Firstly,the spatio-temporal correlation of airport surface traffic was analyzed based on the airport operation data,and the spatio-temporal influence factors of traffic situation were selected based on the spatio-temporal correlation between arrival and departure flights.The airport was modeled as a network diagram,and the flight node time and parking space information were input.The numerical simulation model of airport surface operation was established by MATLAB software,and the position-time information and traffic situation impact factor of each flight were obtained,which provided support for situation prediction.Secondly,the CNN-LSTM prediction model was built,and the influence factors obtained by simulation were converted into space-time feature matrix input prediction model.After extracting the spatial characteristics between the influence factors of airport surface traffic situation by CNN network layer,the time characteristics were extracted by LSTM layer,and the model was trained.After confirming the optimal parameter combination of CNNLSTM prediction model,the departure flight flow,departure flight taxiing time,queuing length and airport surface traffic density of Zhengzhou Airport were predicted.The mean absolute error were calculated to be 0.730,0.739,1.010 and 3.653 respectively.Finally,the CNN-LSTM prediction model was compared with the prediction results of the LSTM prediction model,the BP neural network prediction model,and the Elman neural network prediction model to verify the effectiveness of the situation impact factor and the prediction performance of the CNN-LSTM prediction model.After that,the entropy weight method was used to calculate the weight of the four traffic situation characterization indicators obtained by the prediction,and the quantitative analysis of the airport surface traffic situation prediction value was realized.The experimental results demonstrated that the proposed method for predicting airport surface traffic situations based on spatio-temporal features effectively captured information regarding the spatio-temporal characteristics of airport surface traffic and exhibited strong prediction performance.This method could be applied to predict airport surface traffic situations,thereby providing decision-making support for airport surface traffic control.