Research On Decision-making Algorithm Based On Multiple Tasks Of Multiple UAVs

Whenever a natural disaster occurs,it will be accompanied by emergency rescue.Due to the complexity of the geological structure in the area where the disaster occurred,there will be a continuing danger after the disaster.However,it is often difficult for rescuers to arrive at the scene in the first place.At this time,with its unique characteristics,drones play a key role in the early stage of disaster relief.For example,drones can automatically collect disaster information and distribute emergency supplies.So how to realize the rapid collection of information and the distribution of materials under the condition of uncertain information in the disaster area is a key issue.This article focuses on the complex situation of emergency rescue of drones in the aftermath of a disaster,and solves the problems of multiple drones investigating all potentially affected areas at the same time,and promptly delivering materials in a timely manner.In addition,when performing unmanned aerial vehicle missions,multi-UAV 3D trajectory planning and dynamic avoidance of threat sources are also issues that must be considered.At present,the commonly used multi-UAV multitarget reconnaissance problem model is mainly to solve the problem by optimizing the UAV flight path to obtain the shortest path.On the issue of emergency supplies allocation,the allocation of resources under exact demand is usually considered.This article is based on modeling the time urgency of the reconnaissance mission in a disaster environment and the balance of multiple UAV missions,and uses genetic algorithms to solve the optimization problem of UAV reconnaissance and material allocation.Aiming at the task of drone reconnaissance to the disaster-stricken point,this paper designs a centralized reconnaissance strategy with multiple traveling salesman problem and each traveling salesman task balance.However,when there are too many disaster-hit points,the algorithm complexity is high.In order to simplify this problem,this paper uses the K-means algorithm to divide the affected areas.In the scenario of unmanned aerial vehicle emergency material distribution,due to the uncertain demand of each disaster site,this paper refers to and optimizes the vehicle routing problem model with random demand.Combined with the average distance method,the stochastic demand routing problem of multiple drones and multiple bases is reduced to the stochastic demand routing problem of single drones and single bases.Next,the improved genetic algorithm is used to solve the model.In the three-dimensional multi-UAV trajectory planning and decisionmaking scenario,a functional model is established by establishing the digital terrain of mountains and threat sources.Finally,the simulated annealing particle swarm hybrid algorithm was used to simulate the flight path of the multi-UAV threedimensional environment.This article further discusses the problem of dynamic obstacle avoidance.Compared with the original research,the simulation results of this paper have made some improvements.In the UAV reconnaissance module,the area division greatly reduces the complexity of the algorithm.The solution of the multi-traveling salesman model of task balance not only reduces the difference between the longest path and the shortest path between single drones,but also reduces the task completion time.In the emergency material distribution module of the UAV under the condition of uncertain disaster,this paper verifies the rationality of the UAV’s material distribution through simulation.Finally,in the trajectory planning module,this paper uses the simulated annealing particle swarm hybrid algorithm to solve the path optimization problem in the multi-machine trajectory planning problem.In summary,the rescue plan of the drone proposed in this paper can solve the problem of drone rescue and material distribution under conditions.