基于多智能体的巡逻任务分配和路径优化方法研究

Research on Patrol Task Allocation and Path Optimization Based on Aulti-agent

合理有效的巡逻任务规划对于提高街面"见警率"和巡逻力度、加强社会面动态防控工作具有重要意义,由于巡逻任务规划具有多任务、多约束、多目标等特点,极大增加了建立任务分配算法的难度。面向协同巡逻中任务分配和路径优化问题,提出了一种集成的仿真优化框架,采用基于仿真的方法,将进化算法与优化实验相结合进行协同优化。针对巡逻的人机协同以及选址问题,追求任务分配和路径优化综合代价最小建立二层优化模型,结合多智能体遗传算法和Opt Quest优化实验构建巡逻方案,设计案例验证模型,同时实现规划结果的GIS可视化。仿真结果验证了模型和算法的有效性,该方法具有较强可扩展性,可解决多巡逻和多地址环境下的多目标优化调度问题。

Reasonable and effective patrol task planning is of great significance to improve the "police presence rate"and patrol strength of the street,and to strengthen the dynamic prevention and control of social aspects,As the patrol task planning has the characteristics of multi-task,multi-constraint and multi-objective,it greatly increases the difficulty of establishing the task assignment algorithm. Aiming at task assignment and path optimization in cooperative patrol,an integrated simulation optimization framework was proposed,which combined evolutionary algorithm and optimization experiment for collaborative optimization using simulation-based method. Aiming at man-machine collaboration and location problems of patrol,a two-layer optimization model was established to minimize the comprehensive cost of task allocation and path optimization,and a patrol scheme was constructed by combining multi-agent genetic algorithm and OptQuest optimization experiment. A case was designed to verify the model,and GIS visualization of planning results was realized at the same time. The simulation results verify the effectiveness of the model and algorithm,and the proposed method has strong scalability,which can solve the multi-objective optimization scheduling problem in the multi-patrol and multi-address environment.