Research On Multi-UAV Path Planning Based On Heterogeneous Grouped Particle Swarm Algorithm

With the development of artificial intelligence,autonomous driving and other high technologies in recent years,UAVs as an important role in the modern battlefield is flourishing worldwide in the research and application.A reasonable and safe flight path for UAVs determine directly success of UAV missions.With the complexity of UAV missions,a single UAV can no longer meet the demand,and multi-UAV path planning has emerged.The UAV path planning problem is considered as a multi-constraint combined optimization problem,featuring muti-constraints and high-dimensions.Traditional algorithms are difficult to solve high-dimensional multi-objective problems,resulting in the planned path poorly applied in the real-world problems.Particle swarm algorithm,as a swarm intelligent optimization algorithm,has attracted widely attentions in the UAV path planning problem due to its excellent performance.This work focuses on the model and algorithm of the UAV path planning.Based on the multi-UAV path planning models,two improved particle swarm algorithms have been proposed for the UAV path planning.The main research contents and results in this work are as follows:(1)The establishment of multi-UAV static path planning model mainly includes four parts: coordinate system transformation,track planning terrain modeling,constraint analysis modeling and the construction of comprehensive cost function.Coordinate system transformation connects the start point and end point,selects a new coordinate system,and then simplifies the calculation.In the three-dimensional case,this paper selects the grid digital elevation model for terrain modeling,which mainly realizes the modeling of threats such as mountains.Threat modeling refers to the establishment of threat models such as radar and air defense artillery.The constraint modeling including minimum step constraint,maximum turning angle constraint,maximum pitch angle constraint,speed constraint and so on.The comprehensive cost function includes fuel consumption cost,height cost,threat cost and multi-UAV cooperation cost.(2)A Modified Heterogeneous Comprehensive Learning Particle Swarm Optimization(MHCLPSO)is proposed: firstly,the particle swarm algorithm is balanced by dividing the whole population into two heterogeneous subpopulations named exploration and exploitation.The exploration subpopulation adopts a comprehensive learning strategy to learn only from its own optimal position.The exploitation subpopulation learns not only from its own optimal position,but also from the global optimal position.In order to improve the local search ability of particles and prevent them from falling into local optimum,by borrowing the idea of niching,each particle learns from its own optimal position of multiple individuals in the neighborhood.Meanwhile,an adaptive acceleration constants strategy is incorporated to better facilitate the evolution of the algorithm toward the location of the optimal solution.By comparing with four typical particle swarm algorithms,the proposed MHCLPSO algorithm has the fastest convergence speed and the smallest fitness value,and the planned path is more efficient.(3)For the multi-UAV path planning problem in 3D scenes,this paper proposes a Heterogeneous Adaptive Comprehensive Learning and Dynamic Multi-Swarm Particle Swarm Optimizer(HACLDMS-PSO)to solve this problem.The algorithm integrates three strategies: the population dynamic adjustment strategy,two perturbation strategies,Levy flight and Cauchy mutation,and an adaptive learning probability mechanism.The population dynamic adjustment strategy balances the global search and local search ability of the population.The perturbation mechanism is used to encourage particles to jump out of the local optimal position.The adaptive learning probability mechanism to promote the evolution of particles toward the global optimum.To improve the quality of planning,a constraint handling mechanism based on a comparison criterion is used to handle the constraints.Finally,through simulation experiments and comparative analysis,the superiority of the improved particle swarm algorithm in solving the multiUAV path planning problem is verified.