基于控制力矩陀螺的水下运载器动力学与仿真

为弥补水下运载器(AUV,Autonomous Underwater Vehicle)中传统舵面控制机构的低速控制的不足,改善其操纵性能,引入单框架控制力矩陀螺(SGCMG,Single Gimbal Control Moment Gyro)作为控制机构进行姿态稳定与控制.把AUV简化为刚体,加入SGCMG,考虑水下环境的特点,建立基于SGCMG的AUV动力学模型,并仿真分析AUV的动力学、姿态运动、SGCMG的框架运动以及环境之间的相互作用.仿真结果说明:基于SGCMG控制的AUV的姿态机动快速、准确,低速性能理想,为操纵律设计及姿态控制算法研究提供基础.     

基于Fabry-Perot干涉仪的闭环微光机电加速度计

设计了一种闭环反馈差动式双FP腔的微光机电(MOEMS)加速度计,介绍了其工作原理及系统构成.利用惯性敏感单元将对载体加速度的测量转变为对载体位移的测量,利用光纤自聚焦透镜的端面与质量块组成的FP腔测量载体位移.为了提高系统的测量灵敏度和抑制温度等环境因素的影响,设计了一种差动式双FP腔测量机构.为提高微加速度计的输出线性度和动态测量范围,提出了采用静电力平衡技术构成闭环加速度计.建立了其数学模型,对所设计的加速度计重要参数指标——灵敏度、敏感头受载、固有频率等一一进行了详细计算和分析.在此基础上完成了设计背景要求下加速度计参数的优化设计,结果表明:该系统精度可以达到5×10-6g以上.      

基于改进YOLO V5的无人机航拍图像小目标检测方法

本文针对无人机航拍目标检测技术中目标聚集、目标较小及实时性差等问题,将YOLO V5的主干架构进行改进,简化Neck网络,提出了一种提高检测速度又能准确识别的无人机航拍图像检测技术方案。经过仿真实验测试,改进后的YOLO V5网络在保持识别精度的同时,检测速率提升了31%,满足无人机在航拍作业时对于准确性与实时性的要求。     

光纤陀螺温度建模及补偿技术研究

研究了干涉式光纤陀螺的温度补偿问题。首先分析了光纤陀螺仪温度漂移的主要影响因素，并应用人工神经网络与数理统计方法建立了陀螺仪的温度补偿模型。最后对某型号陀螺进行了0～50℃环境温度下的大量恒温及变温实验，通过实验数据完善所建模型并完成了对该陀螺的温度补偿。实验结果表明：在特定温度环境下，该方法能够有效地改善光纤陀螺仪零偏稳定性。     

基于变速积分的高速电机磁悬浮控制系统设计

针对磁悬浮高速电机的磁轴承-转子系统,设计了基于MCU+FPGA的数字控制系统。为实现转子系统的高转速下稳定控制,在建立了磁轴承-转子闭环控制系统模型的基础上,针对常规PID控制系统的超调振荡问题和PD控制系统的转子位移静差问题,利用变速积分的思想设计了改进的PID控制器。实验表明,在消除了系统静差的同时明显减小了系统的超调量和调节时间,说明设计的控制系统具有很好的动态特性和稳态精度。     

What is wrong with space system cost models? A survey and assessment of cost estimating approaches

Despite several decades of research and refinement in cost estimating tools and practices, the final cost of US space programs continues to exceed initial cost estimates by an average of more than 45%. Thus, program cost models not only exhibit error, they are seriously biased. A structured review surveyed techniques, approaches, models and conceptual tools related to space program cost estimating, to understand cost in complex space systems. Analysis shows problems of cost estimating result from the growing complexity of space programs, failures in managing growth, and mission failures. Although there is greater expectation for the models to accurately predict program costs, the current models used for seeking funding for large space programs are inadequate due to (1) inability to predict future, (2) lack of insight, and (3) process replaces judgment in decision making.     

Return to Venus of the Japanese Venus Climate Orbiter AKATSUKI

Japanese Venus Climate Orbiter/AKATSUKI was proposed in 2001 with strong support by international Venus science community and approved as an ISAS (The Institute of Space and Astronautical Science) mission soon after the proposal. The mission life we expected was more than two Earth years in Venus orbit. AKATSUKI was successfully launched at 06:58:22JST on May 21, 2010, by H-IIA F17. After the separation from H-IIA, the telemetry from AKATSUKI was normally detected by DSN Goldstone station (10:00JST) and the solar cell paddles’ deployment was confirmed. After a successful cruise, the malfunction happened on the propulsion system during the Venus orbit insertion (VOI) on Dec. 7, 2010. The engine shut down before the planned reduction in speed to achieve. The spacecraft did not enter the Venus orbit but entered an orbit around the Sun with a period of 203 days. Most of the fuel still had remained, but the orbital maneuvering engine was found to be broken and unusable. However, we have found an alternate way of achieving orbit by using only the reaction control system (RSC). We had adopted the alternate way for orbital maneuver and three minor maneuvers in Nov. 2011 were successfully done so that AKATSUKI would meet Venus in 2015. We are considering several scenarios for VOI using only RCS.     

SETI implications of gravitational assist via chaotic trajectories of binary objects

Extrapolating from the technique of gravitational assist via chaotic trajectories of binary objects, this paper considers how such techniques might be used in other systems. We examine which types of systems are the best candidates for harvesting gravitational energy for payload ejection. We also consider what signatures might be present in either the asteroid orbits or radiation of the central body if extraterrestrial intelligences were to use such techniques about these candidate systems.The simulation studies show that current technology cannot approach the sensitivity needed to detect either of these signals. Instead, we provide these results as guidance to studies in coming decades on patterns that may indicate the use of an asteroid ejection system.     

Urban growth assessment around Winam Gulf of Kenya based on satellite imagery

Urban growth and population dynamics are among the most critical information needed for future economic development planning, natural resources allocation and environmental management. In the present work, two methods, the first based on night-time images produced by NOAA and population maps provided by Oak Ridge National Laboratory's (ORNL) LandScan, and the second one on SAR imagery, were used in order to assess the expansion of urban areas surrounding the Winam Gulf (Lake Victoria, Kenya) at different scales. In the time covered by night-time lights imagery, the study highlighted a period of constant growth rate between 2002 and 2006 and a negative trend after 2006 and 2008. This decrease may be related to two main events occurring in the study area between 2006 and 2007: the decline of the Lake Victoria level and the abnormal proliferation of the floating weeds within the Winam Gulf. Meanwhile, the urban feature extraction obtained at a different scale within a particular district from 1997 up to 2008 results in a constant growth rate. Population movements around this zone explain different dynamics that should be studied in detail in order to understand their particular roots.     

无人集群系统行为决策学习奖励机制

未来作战的发展方向是由多智能体系统构成的无人集群系统通过智能体之间自主协同来完成作战任务。由于每个智能体自主采取行为和改变状态，增加了智能群体行为策略训练的不稳定性。通过先验约束条件和智能体间的同构特性增强奖励信号的实时性，提高训练效率和学习的稳定性。采用动作空间边界碰撞惩罚、智能体间时空距离约束满足程度奖励；通过智能体在群体中的关系特性，增加智能体之间经验共享，进一步优化学习效率。在实验中，将先验增强的奖励机制和经验共享应用到多智能体深度确定性策略梯度（MADDPG）算法中验证其有效性。结果表明，学习收敛性和稳定性有大幅提高，从而提升了无人集群系统行为学习效率。