Deep reinforcement learning for six degree-of-freedom planetary landing

This work develops a deep reinforcement learning based approach for Six Degree-of-Freedom (DOF) planetary powered descent and landing. Future Mars missions will require advanced guidance, navigation, and control algorithms for the powered descent phase to target specific surface locations and achieve pinpoint accuracy (landing error ellipse <5 m radius). This requires both a navigation system capable of estimating the lander’s state in real-time and a guidance and control system that can map the estimated lander state to a commanded thrust for each lander engine. In this paper, we present a novel integrated guidance and control algorithm designed by applying the principles of reinforcement learning theory. The latter is used to learn a policy mapping the lander’s estimated state directly to a commanded thrust for each engine, resulting in accurate and almost fuel-optimal trajectories over a realistic deployment ellipse. Specifically, we use proximal policy optimization, a policy gradient method, to learn the policy. Another contribution of this paper is the use of different discount rates for terminal and shaping rewards, which significantly enhances optimization performance. We present simulation results demonstrating the guidance and control system’s performance in a 6-DOF simulation environment and demonstrate robustness to noise and system parameter uncertainty.     

论领导干部的表达力

领导干部的表达力,不仅是个人的人力资本,更是组织的宝贵财富。其现实表现具有策略性。通过个人的学习和组织的文化引领,可以实现角色转换、认同、超越,进而改善和提高表达力。     

面向卫星电源系统的一种新颖异常检测方法

面向卫星电源高维周期性时序遥测数据，提出了一种新颖的代表性特征自编码器（RFAE）模型，并用于无监督的异常检测。RFAE采用改进的堆叠自编码器损失函数和训练算法，从而使模型可以学习到相位相同样本的代表性特征；然后根据代表性特征重构样本，根据重构误差来判断样本是否异常。在试验部分首先通过模拟数据校验了RFAE算法能够有效地检测出高维周期性时序数据的异常，然后又采用某卫星电源系统2014年1~12月真实遥测数据进行试验，RFAE异常检测准确率达到99%，检测效果明显优于目前的其他异常检测算法，具有较高应用价值。     

Feasibility analysis of LEO and GEO large space debris de/re-orbiting taking into account launch mass of spacecraft-collector and its configuration layout

Analysis of the efficiency of two basic strategies for de/re-orbiting large space debris objects to disposal orbits (DO) is given. Large objects in LEO are classified into groups with similar orbital inclinations and comprise primarily last stages of launch vehicles, in GEO vicinity the paper studies upper stages. Under the first de/re-orbiting variant, it is assumed a spacecraft-collector is equipped with several thruster de/re-orbiting kits (TDKs); one of them can be fixed on an object and is capable of de/re-orbiting an object to a DO independently of the collector. In the second variant, a collector operates as a space tug: transfers objects to a DO and then returns to the next objects in line. The authors study possible configuration layouts of collectors in LEO and near GEO. The available analogous projects are analyzed. The efficiency of both de/re-orbiting variants can be properly compared using the estimations of collector's dry mass and having at one's disposal the parameters of the maneuvers required for transfers between all objects in the group. As reasonable criteria of effectiveness, one can consider (separately or jointly) the launch mass of an equipped collector, its ΔV budget, and the required number of such active spacecraft. Two de/re-orbiting variants are compared in terms of these criteria via mass-energy diagrams constructed for each group of objects in both altitude regions. Analysis of these diagrams shows that low Earth orbits can be more efficiently cleaned under the first de-orbiting variant by using a two-stage space system consisting of an active spacecraft carrying TDKs. For GEO, it is expedient to choose the second re-orbiting variant using a single-stage spacecraft. Our analysis shows that LEO cleaning is an order of magnitude more expensive than that for GEO, hence the problem of LEO population should be given increased attention.     

基于局部线性嵌入的高光谱影像特征提取算法

特征提取能够消除冗余信息,提高高光谱数据处理的精度和计算效率,是分类等分析必要的预处理手段.传统特征提取算法基于线性变换,无法准确描述高、低维特征空间的关系,因此采用一种新型非线性特征提取算法,即局部线性嵌入(LLE,Locally Linear Em-bedding),挖掘高光谱影像的本征信息.针对分类问题,使用训练样本类别属性修正距离矩阵,并借鉴LLE计算未知样本低维映射的方法求解测试样本的特征向量,实现监督局部线性嵌入(SLLE,Supervised Locally Linear Embedding).使用机载可见光/红外成像光谱仪数据,与3种分类算法结合进行测试,实验结果表明:SLLE优于线性特征提取算法,能够解决高光谱影像的小样本分类问题.     

基于全局信息的粒子群算法翼型综合优化设计

 翼型优化往往需要考虑众多的设计目标和约束条件,对此发展了稳健高效的翼型综合优化方法。在粒子群优化算法中用繁殖策略深度挖掘由Kriging代理模型所获取的全局信息,对基准函数优化、翼型几何外形重构与层流翼型优化问题进行了测试,结果表明该算法可大幅度提高优化速度。将改进的Hicks-Henne翼型参数化方法和雷诺平均Navier-Stokes(N-S)方程流场求解器与优化算法相结合,采用可方便确定权重系数的多目标非线性适应值加权方法,分别对多点、多目标和多约束的超临界翼型与低速翼型进行综合优化,计算结果表明该方法可大大提高气动外形优化的工程实用性。     

基于支持向量机的飞行控制系统辨识

以飞机为研究背景,基于系统辨识中的串-并联辨识模型,应用一种支持向量回归机的方法,对飞行控制系统模型进行了辨识。通过仿真,成功地得到了飞行控制系统的辨识模型。结果表明:该方法有很好的预测性能,具有很高的精度,同时也能很好地反映飞行控制系统的特性。     

基于IRI背景场的单站电离层TEC地图重构技术

为了有效解决电离层TEC观测数据稀疏时重构问题, 引入IRI-2007作为背景场, 利用反距离加权法和克里格方法重构了电离层TEC地图, 使用交叉检验方法检验了引入背景场前后的重构精度. 结果表明, 引入背景场后, 一方面有效地控制了边缘地区的发散现象, 另一方面重构网格点上绝对误差在 -0.25~0.25 TECU之间的比例分别提高了约70 %和100 %, 误差统计基本呈正态分布. 可以通过引入更加精确的背景场或使用逐步订正方法进一步提高重构精度.      

Structured sparsity assisted online convolution sparse coding and its application on weak signature detection

Due to the strong background noise and the acquisition system noise, the useful characteristics are often difficult to be detected. To solve this problem, sparse coding captures a concise representation of the high-level features in the signal using the underlying structure of the signal. Recently, an Online Convolutional Sparse Coding(OCSC) denoising algorithm has been proposed. However, it does not consider the structural characteristics of the signal, the sparsity of each iteration is not eno...     

基于深度强化学习的固定翼无人机编队协调控制方法

由于运动学的复杂性和环境的动态性,控制一组无人机遂行任务目前仍面临较大挑战。首先,以固定翼无人机为研究对象,考虑复杂动态环境的随机性和不确定性,提出了基于无模型深度强化学习的无人机编队协调控制方法。然后,为平衡探索和利用,将ε-greedy策略与模仿策略相结合,提出了ε-imitation动作选择策略;结合双重Q学习和竞争架构对DQN（Deep Q-Network）算法进行改进,提出了ID3QN（Imitative Dueling Double Deep Q-Network）算法以提高算法的学习效率。最后,构建高保真半实物仿真系统进行硬件在环仿真飞行实验,验证了所提算法的适应性和实用性。