Research On Flight Conflict Identification And Deployment Method Based On Neural Network

Activities of flight conflict identification and deployment account for a large proportion in the actual control process.It can reduce the workload and enhance the service capability of air traffic control by studying this activity and developing the corresponding auxiliary decisionmaking system.In this paper,a conflict identification and deployment model based on neural network is proposed.By learning historical data,the control strategy is obtained,and then reliable flight identification and deployment functions are provided and the corresponding system module programs are written.By exploring the relevant theoretical knowledge,the research of this technology theory is often combined with the kinematics of the aircraft to establish the corresponding mathematical model,and the conflict identification can still obtain relatively good results.However,the deployment program is still at the theoretical stage,and it is difficult to transform into a system that can be applied in practice.Therefore,this paper continues the idea of technology transformation,and mainly does the following work:(1)Flight data information processing.Combined with the improved flight protection zone,the potential conflict data are collected;According to the corresponding deployment control instructions of potential conflicts,the deployment data of potential conflicts are summarized;The track history data of the identified potential conflict aircraft group are screened out,and the potential conflict deployment track data are collected.Combined with the track characteristics,the flight data information is preprocessed such as necessary analysis,deduplication,interpolation and alignment.(2)The process of conflict identification,control and deployment were analyzed,and a suitable aircraft flight protection zone was established.Combined with the special buffer,a flight conflict identification model based on improved BP neural network was proposed,which can distinguish the conflict warning time and find potential flight conflicts in advance.The rough set theory and particle swarm optimization were used to optimize the model.(3)Combining the control deployment method with the actual control instruction data,a flight conflict deployment model based on improved BP neural network was proposed.Aiming at the problem that some data have no corresponding control instructions,an anomaly detection method was introduced.(4)From the perspective of track data change in the process of conflict allocation,a conflict allocation model based on LSTM and CNN-LSTM neural network was proposed in combination with the time series and spatio-temporal correlation characteristics of tracks.(5)The flight conflict identification and deployment module was integrated in the simulated air traffic control system,and the effect of the algorithm was verified by typical conflict scenarios.Compared with the traditional algorithm,the deployment trajectory was closer to the strategy used by the controller.Based on historical data,this paper uses neural network algorithm to complete the whole process of flight conflict identification and deployment through the steps of identifying flight conflicts and issuing corresponding deployment programs,which provides a feasible method for neural network algorithm to participate in the auxiliary decision-making in the field of automatic air traffic control in the future.