Research On Key Technologies Of UAV Path Planning Based On Intelligent Optimization Algorithm

With rapid development and increasingly widespread application of Unmanned Aerial Vehicle(UAV)technology,autonomous execution of tasks by UAVs in complex scenarios has become a current research hotspot.Target threat assessment and UAV path planning are important prerequisites and key technologies of UAV task execution.Currently,traditional target threat assessment methods tend to rely on expert knowledge when determining indicator weights,and there are urgent needs to address the shortcomings of suboptimal paths and lack of consistency when using intelligent optimization algorithms for path planning.This thesis focuses on key issues related to UAV path planning,conducting research on problem analysis,model establishment,method design and simulation experiments.By improving dragonfly algorithm(DA)and promoting its application in target threat assessment and UAV path planning,this thesis has important scientific value for ensuring path safety,shortening path length,expanding UAV application scenarios and improving the level of intelligence in UAV technology autonomy.The main work and innovative points of this thesis include the following 3 aspects:(1)Tent Chaotic Map and Population Classification Evolution Strategy-Based Dragonfly Algorithm(TPDA)is proposed.To address the shortcomings of slow convergence speed,low accuracy,and easy trapping in local optima in optimizing functions using DA,patterns of population initialization and algorithm iteration are improved.Tent chaotic map is used to generate a more uniformly distributed initial solution population,ensuring that the algorithm has a good iteration starting point;According to the characteristics of DA,the population is classified based on fitness value and different location update methods are used for individuals with different fitness values in combination with the historical optimal location of individuals and populations,so as to improve the ability to jump out of local optima for global optimization.18 10-dimension functions and 14 30-dimension functions are selected for function optimization using 4 algorithms:TPDA,PSO,DA,and ADDA.Results show that in terms of average values,TPDA outperforms the others in all 18 and 14 functions respectively;In terms of standard deviation,TPDA performs better on 15 and 13 functions,accounting for 83.3% and 92.9% respectively.TPDA has better optimization accuracy and stability than other algorithms in terms of high dimension,multi extreme value and nonlinear function,and has good application prospects.(2)A target threat assessment method based on the fusion of TPDA and projection pursuit model is proposed.In response to the limitation of traditional threat assessment methods relying on expert knowledge when determining indicator weights,the size of indicator weights is determined through projection pursuit model,thereby overcoming the subjective dependence of traditional methods manually calibrating indicator weights.The original data is projected into a low dimensional space,construct an objective function and the objective function to be optimized is constructed through the inter class dispersion and intra class density of the projected eigenvalues.Furthermore,in response to the difficulty of traditional algorithms in optimizing the objective function which makes it difficult to converge to the global optima,TPDA is used to obtain the optimal weights of each indicator,the threat degree of targets and ranking results.3 typical examples were selected and simulated using TPDA and PSO.Results show that TPDA can achieve consistent target threat ranking results with the original text on all instances,while PSO cannot distinguish the degree of target threat on 1 instance,resulting in evaluation failure.The target threat assessment method based on the fusion of TPDA algorithm and projection pursuit model can mine the inherent laws of data,reflect the relative importance of indicators,and efficiently and stably calculate the optimal indicator weight compared to other algorithms.It has strong migration ability and can quickly conduct threat assessment in dynamic changing scenarios,meeting the requirements of target threat assessment in future battlefield environments.(3)A UAV path planning method based on TPDA has been proposed,meeting the requirements of migration applicability between different maps and multi path point generation.In response to the shortcomings of traditional path planning methods that rely on the construction of planning maps and the direct proportion of planning time to the square of the number of path points,a universally applicable drone path planning map is constructed.The path and number of path points are treated as dragonflies and dimensions in TPDA and a collision detection method between path and obstacles is designed to establish an objective function to be optimized.TPDA and PSO are used to carry out UAV path planning simulation experiments.Results show that TPDA is significantly better than PSO in the 2D and 3D UAV path planning,and with the increase of the number of track nodes and the complexity of map space,the performance and robustness of the TPDA are more prominent: in 3D space,there is an obvious “detour” phenomenon in the tracks planned by PSO,10% and 54% longer than path planned by TPDA when generating 5 and 10 path nodes;When conducting path planning on complex maps,the path obtained by TPDA is at least30% shorter than that obtained by PSO.The path planning method proposed in this article can plan safer and more efficient flight paths in complex environments,with stronger robustness and can meet the needs of future unmanned aerial vehicle path planning.