Flight Conflict Resolution Method Based On Controller Habits Learning

With the rapid development of social economy,the volume of civil aviation transportation is also growing rapidly.How to ensure safe and efficient civil aviation transportation within limited airspace resources is the primary task of air traffic management.Conflict detection and resolution is an important focus of safety assurance.In order to reduce the interference of human factors in the process of generating the conflict resolutions,it is a general trend to introduce intelligent systems to assist air traffic controllers.However,due to the restrictions of the controller’s habits and knowledge level,the solutions given by the intelligent system are often not understood and trusted by the controller and eventually abandoned.This research uses machine learning algorithms to learn the working habits of air traffic controllers on duty,uses digitization as an influencing factor in the process of generating flight conflict resolution solutions,and focuses on the method of generating customized solutions.First,based on the route information provided by the flight plan,4D trajectory modeling was performed on the predicted trajectory,and based on the 4D trajectory model,the potential flight conflicts between routes were detected through the detection algorithm of time and space.For conflicting routes,generating conflict resolution solutions based on speed adjustment or heading adjustment,and use simulated annealing algorithms to optimize the solutions,finally provide optimal conflict resolution solutions under different types.Subsequently,the machine learning Q-learning algorithm is used to deal with the potential conflict resolution data with less state changes and more action items to learn the habits of controllers,and includes the consideration of the effectiveness of the actions of the controllers,giving numerical controller habit preference data.Finally,combined with traditional flight conflict detection and resolution methods,a customized conflict demodulation method optimization scheme that fits the preferences of controllers is constructed.Similarly,the simulated annealing algorithm is used as an optimization method.When the preference parameters representing the habits of controllers are introduced into the objective function,the simulated annealing multi-objective optimization method is used to solve the multi-objective optimization problem that needs to consider the preferences of the controller and the parameters that affect the original route.Meet the actual application scenarios of the controller.The customized algorithm also gives a customized solution that is more suitable for specific controllers than the traditional optimal solution.Finally,the simulation experiment and performance analysis of the algorithm are carried out.From the influence of the introduction of controller habit parameters on the type of release plan and the changes of various parameters in the release plan,the influence of the introduction of controller habit parameters on the release plan and the degree of compliance with the preference of the controller are analyzed from multiple perspectives,and a customized plan is obtained.The solution given without affecting the ability of conflict demodulation is easier for the controller to understand and use,and reduces the controller’s work pressure.