面向强干扰对抗的无人艇智能识别算法鲁棒性测试评估方法
Robustness Test and Evaluation Method of Unmanned Vessel Intelligent Recognition Algorithm for Strong Interference Confrontation
宋胜男 1罗荣 2白高颐 1杨星 3陈卓4
作者信息
- 1. 中船智海创新研究院,北京 100036
- 2. 91054 部队,北京 102442
- 3. 31453 部队,北京 110023
- 4. 中船智海创新研究院,北京 100036;浙江大学,杭州 310058
- 折叠
摘要
针对水面无人艇在实际环境中面临的海上复杂环境干扰和人为攻击,为提高智能识别算法的可靠性和安全性,本文提出了一种面向强干扰对抗的无人艇智能识别算法鲁棒性测试评估方法.该方法主要包括无人艇智能识别算法鲁棒性评估的指标体系、测试评估流程、模拟海上海况和天气条件干扰,以及人为算法攻击的测试数据集生成方法.通过仿真验证结果,证实了该测试评估方法的有效性,为可靠性测试提供更全面和准确的评估.
Abstract
In order to improve the reliability and safety of the intelligent recognition algorithm for the complex environmental interference and human attack at sea faced by unmanned vessel in the actual environment,this study proposes a robustness test and an evaluation method for unmanned vessel intelligent recognition algorithm for the strong interference confrontation,which mainly includes the index system of the unmanned vessel intelligent recognition algorithm robustness evaluation,test and evalua-tion process,the method of generating test data sets for simulation of the interference of the sea state and weather conditions at sea and the attack of man-made algorithms.The validity of the test and evaluation method is confirmed by the simulation verification results,which provides a more comprehensive and accurate assessment for reliability test.
关键词
水面无人艇/智能识别算法/鲁棒性/测试评估方法Key words
unmanned vessel/intelligent recognition algorithms/robustness/test and evaluation methods引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据