Research And Modeling Of Flight Support Based On Bayesian Networks

The network structure of portraying the flight ground support process is the primary problem for predicting the start time of each link and getting the time for accurate time of off-block.Due to the impact of airport support resources and actual operation conditions,the network structure based on the plan support process cannot truly reflect the flight support process.To this end,it is proposed to use the Bayesian structure learning method to learn the flight support structure from the data.In order to accurately describe the network structure of the flight support process determined from the data,by analyzing the characteristics of the flight support service process,a method of portraying the flight support process using the Petri nets of time delay place is proposed.This kind of network associates the time factor with the place,and solves the defect that the change indicates that the network status cannot be represented by the reachable graph.Bayesian networks structure learning is used to obtain the flight support network structure from the support data.Moreover,the network characterization method is improved.Based on the Petri net of the time delay place,the Bayesian network is integrated to obtain the Bayesian network based on the loopless assignment.The Petri net of time delay place is used to describe the results of Bayesian structure learning.In the case of a single flight small data set,Bootstrap is introduced with a sampling method to expand the data set,and then the Bayesian structure is learned by K2 algorithm,and structural learning is performed on each sample,and the edge of each learning result is counted,the edge with higher frequency will be used as the local link structure of the support link that may appear in the network,and the flight support edge frequency network will be obtained.Finally,based on the edge frequency network,a model optimization algorithm is added to search the complete structure of the flight support network by using the ant colony algorithm.The probabilistic transformation rules in the ant colony algorithm are improved,which makes the results of Bootstrap guide the ant colony search algorithm in the process of ant searching for the optimal network structure,and finally obtains the complete flight support network.