With the progress of technology, the multi-agent system is successfully applied in many applications. In this paper, we investigate the problem of multi-agent system reconnaissance mission scheduling, which is the core of the reconnaissance decision support system and can be modeled as an extension of Multi-Mode Multi-Skill Resource-Constrained Project Scheduling Problem. Three objectives are considered in this paper: (1) minimizing the reconnaissance mission’s makespan, (2) minimizing the total cost of allocating reconnaissance agents, and (3) maximizing the total quality of all reconnaissance tasks. An effective problem-specific multi-objective invasive weed optimization algorithm (PS-MOIWO) is proposed for solving the problem. Firstly, a new chromosome structure guaranteeing the feasibility of solutions and an initialization method are proposed. Secondly, we propose a self-adaptive penalty-based constraint handling technique to describe the fitness of each individual and adopt a novel non-dominated sorting method to rank the population. Thirdly, by using the problem-specific knowledge, a local search procedure is developed and incorporated into the PS-MOIWO framework to enhance the exploitation ability. Based on the Taguchi method, algorithm’s suitable parameter combinations are determined. Simulation results based on a set of newly generated reconnaissance instances and the comparisons with some existing algorithms demonstrate the proposed algorithm’s effectiveness.

随着技术的进步，多智能体系统已成功应用于许多应用中。在本文中，我们研究了多智能体系统侦察任务调度问题，它是侦察决策支持系统的核心，可以建模为多模式多技能资源受限项目调度问题的扩展。本文考虑了三个目标：（1）最小化侦察任务的有效期；（2）最小化分配侦察人员的总成本；（3）最大化所有侦察任务的总质量。提出了一种有效的针对特定问题的多目标入侵杂草优化算法（PS-MOIWO）。首先，提出了保证解的可行性的新的染色体结构和初始化方法。其次，我们提出了一种基于自适应惩罚的约束处理技术来描述每个个体的适应度，并采用一种新颖的非支配排序方法对总体进行排序。第三，通过使用特定于问题的知识，开发了本地搜索程序并将其合并到PS-MOIWO框架中，以增强开发能力。基于Taguchi方法，确定算法的合适参数组合。基于一组新生成的侦察实例的仿真结果以及与一些现有算法的比较证明了该算法的有效性。通过使用特定于问题的知识，开发了本地搜索程序，并将其合并到PS-MOIWO框架中以增强开发能力。基于Taguchi方法，确定算法的合适参数组合。基于一组新生成的侦察实例的仿真结果以及与一些现有算法的比较证明了该算法的有效性。通过使用特定于问题的知识，开发了本地搜索程序，并将其合并到PS-MOIWO框架中以增强开发能力。基于Taguchi方法，确定算法的合适参数组合。基于一组新生成的侦察实例的仿真结果以及与一些现有算法的比较证明了该算法的有效性。