Research On Arrival Runway Occupancy Time Prediction Of Aircraft Based On Machine Learning

At this stage,people’s demand for a better life is gradually increasing.In terms of the choice of travel mode,civil aviation is becoming more and more popular.How to promote the improvement of overall operation efficiency under the premise of ensuring safety has become the key.The occupation time of aircraft landing runway and wake interval are two major factors affecting runway capacity,which jointly restrict the improvement of operation efficiency.However,due to the lack of research on the occupation time of landing runway in China,shortening wake interval to improve efficiency has not fully played its due role.Therefore,it is of certain practical significance to carry out extensive research on the occupation time of aircraft landing runway to improve operation efficiency.In this thesis,machine learning method,which is widely used in various fields,is used to predict the occupation time of aircraft landing runway.It is hoped that the prediction model established by obtaining data from typical airports can be used as a reference for the prediction of other airports and help other airports to save some time and financial resources when doing similar research.At the same time,the influencing factors of runway occupancy time are analyzed,which can provide some theoretical support for the study of running-sliding configuration design and optimization of airport planning in China.The main research contents of this thesis are as follows:First,the definition of aircraft landing runway occupancy time(AROT)in this paper is clarified by studying foreign definitions of landing runway occupancy time and China’s operation status.The specific occupation time of each landing stage is further analyzed,and the specific definition of AROT in each occupation stage is explained.Secondly,various factors affecting the occupation time of landing runway are discussed in detail to provide theoretical support for the selection of model characteristic parameters.Then,the characteristics of the data of the monitoring equipment and airborne equipment are discussed,and the target airports for data collection are selected by classification.Further data pretreatment is carried out to prepare the data for the establishment of AROT prediction model.On this basis,the correlation analysis of feature parameters is carried out,the principle of neural network and the structure of random forest in machine learning algorithm are studied,and the AROT prediction model based on BP neural network and AROT prediction model based on random forest are established respectively,and the prediction results are compared and evaluated,and the improvement and optimization direction is proposed.Finally,aiming at the defects of the initial prediction model,a combination of genetic algorithm and particle swarm optimization(GA-PSO)was adopted to improve and optimize the model.The results show that the prediction accuracy of the optimized model is higher than that of the original model,and the accuracy is improved by 15.2%.Furthermore,the influence degree of each characteristic parameter is quantified by SHAP model.