| 引用本文: | 杨静,陈丽,吴金平,陆铭华.经验表征与关系特征联合推理的潜艇防御鱼雷智能决策方法[J].控制理论与应用,2025,42(7):1285~1293.[点击复制] |
| YANG Jing,CHEN Li,WU Jin-ping,LU Ming-hua.Intelligent defense decision-making method by joint reasoning of empirical representation and multi-variable[J].Control Theory & Applications,2025,42(7):1285~1293.[点击复制] |
|
| 经验表征与关系特征联合推理的潜艇防御鱼雷智能决策方法 |
| Intelligent defense decision-making method by joint reasoning of empirical representation and multi-variable |
| 摘要点击 3411 全文点击 247 投稿时间：2023-03-27 修订日期：2024-12-11 |
| 查看全文 查看/发表评论 下载PDF阅读器 HTML |
| DOI编号 10.7641/CTA.2024.30170 |
| 2025,42(7):1285-1293 |
| 中文关键词 潜艇防御 关键信息缺失 经验信息表征 混合属性挖掘 智能联合推理 |
| 英文关键词 underwater defense lack of key information empirical information representation multi-variable intelligent joint reasoning |
| 基金项目 国防科技基础加强计划项目资助. |
|
| 中文摘要 |
| 针对现有战术决策方法人工依赖程度较高、难以有效应对水下不确定决策环境的现状, 本文围绕潜艇防御
来袭鱼雷的智能决策方法进行了研究. 针对被动声呐工作方式对鱼雷距离探测缺失问题, 构建了基于经验表征的缺
失信息表征与预测方法; 同时, 提出通过挖掘对抗系统混合属性的关系, 将其提取为多变量时序分类问题, 构建能够
对抗不确定信息的综合策略选择方法; 结合经验信息表征与关系特征挖掘, 形成综合经验趋势与关系特征的联合
推理智能防御决策框架. 实验结果表明: 所提模型在关键信息缺失条件下决策精度达到了94%, 远高于其他策略选
择方法. 与人在回路线上推演相比, 该框架在无人为干预条件下, 策略选择时间相比于人类指挥员平均快27~36 s,
且防御成功率相比于人在回路提高了8%~16%, 可以有效缓解水下关键缺失的问题, 并在此基础上推广至水下战术
决策的各类场景. |
| 英文摘要 |
| In view of the current situation that the existing tactical decision-making methods are highly dependent on
human reasoning and difficult to deal with the incomplete data underwater effectively, we proposed an intelligent decision-making method for submarine defense torpedoes. Aiming at the problem of missing distance of target caused by passive
working mode of submarines’ sonar, a method of knowledge representation and prediction based on empirical experience
is constructed; At the same time, it proposes to mine the relationship between the multi-variables and extract the hidden
information, so as to build a intelligent classification model that can resist incomplete information; Combined with empirical
knowledge and multi-variables by observing features, a joint reasoning framework integrating empirical knowledge and
relational features is constructed. Experiments show that the proposed method can effectively reduce human intervention
and achieve autonomous defense of torpedoes. Compared with the man-in-the-loop online simulation, the decision time of
proposed method is 27~36 s faster than that of human commanders on average, and the defense success rate is increased by
8%~16% compared with that of man-in-the-loop, which can effectively deal with the missing critical information problems
underwater and can be extended to various scenarios of underwater tactical decision-making in future. |
|
|
|
|
|