摘要
为了识别载人航天任务中手控交会对接人误行为,以《坎巴拉太空计划》为仿真模型构建实验环境,基于机器学习算法构建了一种人误行为分析与识别的方法。通过模拟采集手控交会对接任务过程中操作员的6项生理指标,对各项指标进行机器学习建模分析,筛选出更适用于识别的指标及其最适合的学习器,并将各学习器通过Stacking集成算法进行集成。模型的仿真结果表明,测试集预测精度达到96.36%,验证了上述方法的有效性,为手控交会对接中的人误行为识别分析提供有效的补充与参考。
In order to identify the human errors of rendezvous and docking in manned space missions,this paper build an experimental environment using the Kerbal Space Program as a simulation model,and a method for human error behaviors analysis and recognition is constructed based on machine learning algori thms.By simulating the collection of six physiological indicators of the operator during the rendezvous and docking task,machine learning modeling analysis is performed for eachindicat or to filter out the indicators that are more applicable for identification and the most suitable learners,and the learners are integrated through stacking ensemble alg orithm.The simulation results of the model suggest that the prediction accu racy of the test set reaches 96.36%,which verifies the effectiveness of the method and provides an effective supplement and reference for the analysis of human error behavior identification in rendezvous and docking.
作者
袁成炜
张力
方小勇
刘建桥
YUAN Chen-wei;ZHANG Li;FANG Xiao-yong;LIU Jian-qiao(Computer School/Software School,University of South China,Hengyang Hunan 421200,China;Institute of Human Factors and Safety Engineering,Hunan Institute of Technology,Hengyang Hunan 421200,China)
出处
《计算机仿真》
2024年第7期94-99,151,共7页
Computer Simulation
基金
装备预研国防科技重点实验室基金(614222202040571)
湖南省自然科学基金项目(2015JJ6028)
湖南省教育厅优秀青年基金(21B0798)
湖南工学院科研基金项目(HQ21021)。
关键词
集成学习
交会对接
机器学习
人因失误
模式识别
Ensemble learning
Rendezvous and docking
Machine learning
Human error
Pattern recognition
