Research On Aircraft Surface Taxiing Path Optimization Based On Machine Learning

The rapid development of civil aviation industry has brought a large number of take-off and landing flights and passengers to the civil air transport airport.At the same time,it has affected the operation efficiency of the civil transport airport and increased the workload of airport controllers.The operational efficiency of a large number of transport airports across the country has decreased and the flight delay rate has increased.In order to alleviate the delay of civil transport airport and improve the efficiency of air transport,this thesis optimizes the taxiing path of aircraft on the airport surface.The taxiway,one of the important components of the airport,is studied in combination with the requirements of the 14 th five years plan of civil aviation,so as to improve the operation efficiency of the airport,build a green airport and reduce the operation cost of airlines on the premise of ensuring safety.This thesis studies how to use machine learning to optimize aircraft taxiing path.Firstly,the pavement bearing strength of the selected airport is analyzed to explain the operation limitations of aircraft in the airport.Then,the machine learning model is studied by using Markov decision process,and then the process of path planning by traditional Dijkstra algorithm and Q-learning algorithm is analyzed.Finally,the taxiing path of Wuhan Tianhe International Airport is planned by using machine learning based on Q-learning theory,and compared with the path planned by Dijkstra algorithm based on the Airport.In this thesis,the Markov decision process is used to solve the taxiway path optimization problem.Firstly,the operation constraints of Wuhan Tianhe Airport are added to the research of aircraft path optimization,and the Markov decision problem is simplified by using Bellman equation.Then,the machine learning example is analyzed.By processing the airport structure diagram,the grid environment diagram based on Wuhan Tianhe International Airport is constructed on the basis of the machine learning examples,that is to convert the airport structure map and grid map.In the construction of path planning model using Q-learning algorithm,the path planning is carried out based on Wuhan Tianhe International Airport.The path planning problem of single arrival(departure)aircraft,the path planning problem of single arrival(departure)aircraft based on the temporary closure of taxiway on the surface and the path planning problem of arrival(departure)aircraft based on obstacle avoidance are analyzed respectively.To sum up,this thesis has done a certain amount of research on aircraft surface taxiing path optimization based on machine learning,which has made an important contribution to further improving taxiway capacity,improving airport surface capacity and reducing controller workload.