The multi-target detection and tracking algorithm based on dynamic programming(DP)involves the problem of computational explosion and difficulty in distinguishing adjacent targets.Aiming at the unknown number of targets in practical application and the limitation of strong and weak targets co-existing in environment,this paper proposes a multi-target composite detection and tracking algorithm based on dynamic programming.Firstly,the traditional constant false alarm rate(CFAR)algorithm is used to perform the initial detection so as to identify and eliminate the strong targets.Then,the DP-based track-before-detect(TBD)algorithm is adopted to perform partitioning dynamic programming on the data after the initial detection to realize the detection and track-ing of weak targets,thus achieving the effective detection of dense targets in complex environment.Finally,the performance of the proposed algorithm is verified by simulation and measured data,and the results show that the algorithm is efficient and robust.
为提高现代战争中对飞机目标分类识别的准确度,提出了一种基于卷积神经网络与迁移学习相结合的飞机目标分类方法.首先利用建立的飞机叶片回波模型经过短时傅里叶变换后得到所需直升机、螺旋桨飞机和喷气式飞机的时频图数据集;其次通过迁移学习对源域网络模型进行迁移,并对目标域网络模型通过微调作进一步优化;最后利用所提方法对仿真生成的三类飞机时频图数据集进行预测分类.仿真结果表明,本文方法对三类飞机目标总测试样本的平均分类识别率为99.08%,表现出了良好的分类识别效果.
为提升雷达兵作战指挥课程教学质量效益,分析了课程现状和教学痛点,从优化课程教学内容、创新教学方法、优化组织实施、完善资源条件、改进考核评价五个方面创新探索,拓展了"教"与"学"的维度.教学实践表明,学员对课程教学目标达成和教学质量满意度评价均达到90%以上,连续两年课程考核优良率达到75%以上,课程教学创新取得了较好成效.
为提高舰载预警直升机的警戒巡逻作战效能,需根据典型威胁的目标特性对其阵位配置进行科学规划.首先对舰载预警直升机阵位配置所需考虑的编队威胁角、前出距离、有效预警扇面进行了分析;然后给出了阵位配置需求分析基本流程;最后针对舰基LRASM反舰导弹这一典型威胁对象,进行了仿真计算.研究结果表明:有效探测距离和前出距离是决定舰载预警直升机阵位配置的关键因素;有效预警角随着有效探测距离的增大而增大,当有效探测距离确定时,有效预警角与前出距离呈非线性关系.
为培养高素质的无人机专业人才,分析了我院无人机专业第二课堂教学存在的问题,提出了一种基于成果导向教育(OBE)的无人机专业第二课堂教学方法.该方法从无人机专业的培养目标出发,制定第二课堂的教学目标,逆向设计了以学员为中心的教学实施过程.教学实践表明,该方法的实施提升了学员的实践能力、创新精神和综合素质,取得了良好的成效.
