城市区域多物流无人机协同任务分配

针对城市区域多无人机协同物流任务分配问题, 综合考虑不同无人机性能、物流时效性、飞行可靠性等影响因素, 以经济成本、时间损失和安全风险最小为目标函数, 构建多无人机协同物流任务分配模型。因问题规模大、求解复杂度高, 设计改进的量子粒子群算法进行求解。首先，为增强粒子遍历性和多样性, 采用均匀化级联Logistic映射进行粒子初始化; 其次，为避免算法陷入局部最优解, 引入基于高斯分布的粒子变异方式; 最后，为提高算法运行效率, 运用自适应惯性权重方法对粒子赋值。仿真实验结果表明，所构建的模型能够实现任务分配多目标优化, 贴近城市区域无人机物流配送实际; 所提算法与传统量子粒子群算法和遗传算法相比, 任务分配代价分别下降了5.9%和6.3%;并进一步对参数权重设置进行分析, 当3个子目标函数权重系数分别为0.225、0.275和0.500, 种群规模为150时, 算法规划的结果最优。

Multiple logistics unmanned aerial vehicle collaborative task allocation in urban areas

To solve the problem of multiple unmanned aerial vehicle (UAV) collaborative logistics task allocation in urban areas, a collaborative logistics task allocation model for multiple UAVs is constructed with the objective function of minimizing economic cost, time penalty and safety risk by comprehensively considering the factors of different UAVs'performance, logistics timeliness and flight reliability. Because of the large scale and high complexity of task allocation, an improved quantum particle swarm optimization algorithm is designed to solve the problem. Firstly, in order to enhance the ergodicity and diversity of particles, particle initialization is carried out by homogenization cascade Logistic mapping. Then, in order to improve the search ability of particles and speed up the running efficiency of the algorithm, particle mutation method based on Gaussian distribution is introduced. Finally, in order to improve the efficiency of the algorithm, adaptive inertia weight method is used to assign different values to each particle. Simulation results show that the constructed model can achieve multi-objective optimization of task allocation. It is close to the reality of UAV logistics distribution in urban areas. Compared with the traditional quantum particle swarm optimization algorithm and the genetic algorithm, the task allocation cost of the improved algorithm is reduced by 5.9% and 6.3% respectively. Furthermore, the parameter weight setting is also analyzed in this paper. When the weight coefficients of the three sub objective functions are 0.225, 0.275 and 0.500 respectively and the population size is 150, the task allocation scheme planned by this algorithm is optimal.