针对无人舰艇集群协同搜索中多目标优化与动态约束难以平衡的问题，本文提出多目标约束马尔可夫决策过程算法。首先构建该算法数学模型，同时提出按搜索阶段自适应调整多目标权重的帕累托最优价值函数求解机制。基于Matlab/Simulink-ROS与ADCIRC海洋环流模型构建实验平台，对比传统MDP、分布式一致性、混合整数规划算法。结果表明：本文算法3000步后以0.896高覆盖度收敛；高覆盖度方案任务周期集中于2280–2400 s，通信中断率30%时性能下降幅度较传统MDP低10%以上。本文算法可支撑开阔海域搜救、岛礁海域安防巡逻等场景，兼具理论创新性与工程实用性。
To address the difficulty in balancing multi-objective optimization and dynamic constraints during the collaborative search of unmanned ship clusters, this paper proposes a Multi-Objective Constrained Markov Decision Process (MOC-MDP) algorithm. Firstly, its mathematical model is constructed, and a Pareto optimal value function solution mechanism is proposed—one that adaptively adjusts the weights of multiple objectives according to the search stage.An experimental platform was built based on the Matlab/Simulink-ROS and ADCIRC ocean circulation models to compare the proposed algorithm with traditional Markov Decision Process (MDP), distributed consensus, and mixed integer programming algorithms. Results show that the proposed algorithm converges after 3000 steps, achieving a high coverage rate of 0.896; the task cycle of high-coverage solutions is concentrated between 2280 and 2400 s. When the communication outage rate reaches 30%, the algorithm’s performance degradation is over 10% lower than that of traditional MDP.This algorithm can support scenarios such as open-sea search and rescue and island-reef water security patrols, integrating theoretical innovation with engineering practicality.
2025,47(19): 197-200 收稿日期：2025-4-15
DOI：10.3404/j.issn.1672-7649.2025.19.032
分类号：U662.9; U664.2
作者简介：吴曼丽(1982-),女,硕士,讲师,研究方向为数据库、智能信号与算法仿真测试等
参考文献：
[1] 李正阳, 梁洪瑜, 刘翔宇, 等. 面向舰船受限保障能力的舰基航空器运送任务规划方法[J]. 中国舰船研究, 2025, 10(7): 1-9.
[2] 顾兆军, 赵欢, 王家亮, 等. 基于马尔可夫决策的四轴飞行器自动着陆方法[J]. 航空学报, 2024, 45(15): 302-318.
[3] 李林桐, 蔡育炜, 陆屹, 等. 基于蒙特卡洛树搜索的机载雷达编队策略优化[J]. 飞机设计, 2023, 43(2): 44-48.
[4] 张会, 王安, 程健, 等. 舰船往返拦截搜索发现概率计算公式的推导、检验和修正[J]. 陆军工程大学学报, 2022, 1(1): 54-59.
[5] 朱梦圆, 吕娜, 陈柯帆, 等. 航空集群协同搜索马尔可夫运动目标方法[J]. 系统工程与电子技术, 2019, 41(9): 2041-2047.
[6] 杜永浩, 邢立宁, 陈盈果. 多平台海上协同搜索与路径优化策略研究[J]. 控制与决策, 2020, 35(1): 147-154.
