| In recent years,general aviation in China industry has developed rapidly.However,limited by operational costs,aircraft performance,pilots’ technology and other factors,general aviation flight safety has great risks.Flight efficiency also needs to be improved.Therefore,it is necessary to investigate the method of flight path planning for general aviation to provide decision support for general aviation pilots,so as to improve the safety and efficiency of general aviation.In this thesis,by using advanced and efficient method of reinforcement learning in artificial intelligence,combined with airborne equipment,under the conditions of complex terrain scene,hazardous weather scenario,general flight autonomous path planning method is studied.This thesis uses the Graham method to define the boundary of the hazardous weather and divide geometry of flying in the hazardous weather flight restricted area.The airspace is modeled by the grid method.Q-learning algorithm is used to adjust the reward,learning factor and discount factor in the algorithm according to the action strategy of Soft-max and the actual situation of the airspace.Rerouting path planning is carried out for block flight restricted area,strip flight restricted area and scatter flight restricted area respectively,and the rerouting calculation results are compared with Dijkstra algorithm.The results show that the Q-learning algorithm can obtain the optimal path,and at the same time,the operation time is greatly improved,and the number of turning is also less.At the same time,Q-learning algorithm can generate safe flight path from take-off point to landing point and avoid key obstacles in complex terrain and hazardous weather based on data input from aircraft performance database,airborne terrain database and TCAS system.It can provide decision support for general aviation pilots operating in complex terrain and hazardous weather. |
