基于GA-ELM的飞行载荷参数识别

针对用复杂飞行数据识别飞行载荷时的精度低、速度慢等问题,提出一种结合遗传算法(GA)和极限学习机(ELM)的GA-ELM模型。该模型使用ELM神经网络作为计算核心,用遗传算法产生ELM网络输入层到隐含层的权值矩阵和隐含层偏移量;用GA-ELM模型对飞行数据进行识别,并与BP神经网络和原始ELM神经网络的识别结果进行对比。结果表明:GA-ELM模型是一种有效且高精度的飞行载荷参数识别方法。     

MBD环境下计算机辅助零件分类技术

针对MBD环境下的零件分类问题，研究MBD信息提取、零件编码转换和深度学习分类算法技术，并创建MBD-计算机辅助零件分类系统（MBD-CAPC），实现从输入MBD模型到输出零件分类结果的自动化。     

试析马克思理论体系的两个思想资源

除了常说的三大理论来源以外,马克思的理论体系的形成还有很多丰富的思想资源,其中,基督教和《圣经》传统、古希腊神话和西方文学作品对马克思科学世界观、人生观、价值观的形成产生了深刻影响,马克思对他们采取了批判的借鉴的态度,从中汲取了丰富的营养,成为马克思理论体系重要的思想资源。     

Atmospheric response to a Hot SST Event in November 2006 as observed by the AIRS instrument

Using Atmospheric Infrared Sounder (AIRS) products of atmospheric temperature and geopotential height, we investigate the atmospheric response to HE0611, which was found and investigated by [Qin, H., Kawamura, H., Sakaida, F., Ando, K. A case study of the tropical Hot Event in November 2006 (HE0611) using a geostationary meteorological satellite and the TAO/TRITON mooring array. J. Geophys. Res. 113, C08045, doi: 10.1029/2007JC004640, 2008]. HE0611 was formed by connecting two very high SST areas, HE0611-East and HE0611-West. The period-mean atmosphere temperatures at levels of 925 and 850 hPa in HE0611-West are higher, by about 0.5 K, than those in WE0611-East while the atmospheric temperatures at middle to high levels (700–300 hPa) are higher in HE0611-East. The period-mean geopotential heights HE0611-East are much lower than those in HE0611-West for the levels from the surface to 400 hPa. The mean geopotential heights from 400 hPa to 200 hPa are higher in HE0611-East. In the middle and high layers over HE0611-West, the atmosphere temperatures gradually decrease from 7th to 17th, and then increase significantly. The increase in HE0611-East starts from 15th November, which is earlier than that of HE0611-West. The geopotential heights in the high layer of both the areas also show corresponding behaviors. The lagged atmospheric response in the western part is confirmed by the correlation analysis. It emerges that the atmospheric response to HE0611 is well organized and associated with deep convention in HE0611-East and subsidence in HE0611-West. These are also consistent with the HE0611 features and evolution revealed by earlier HE studies.     

情感因素对语言学习的影响

情感因素对语言学习具有重大的影响。在学生智商水平相当的情况下,情感差异会导致学生学习成绩的高低之分。情感差异是由家庭、环境、情趣、学习方法等综合因素造成的。积极的情感会对学生的学习起到促进作用；反之,消极的情感则会阻碍学生的学习。教师在教学中应善于引导学生,促进学生积极情感的产生,减少消极情感的作用,以提高教学效果。     

基于深度增强学习的卫星姿态控制方法

针对卫星在执行丢弃载荷或捕获目标等复杂任务时遭遇的姿态突然发生变化的问题,采用深度增强学习方法对卫星姿态进行控制,使卫星恢复稳定状态。具体来说,首先搭建飞行器的姿态动力学环境,并将连续的控制力矩输出离散化,然后采用Deep Q Network算法进行卫星自主姿态控制训练,以姿态角速度趋于稳定作为奖励获得离散行为的最优智能输出。仿真试验表明,面向空间卫星姿态控制的深度增强学习算法能够在卫星受到突发随机扰动后稳定卫星姿态,并能有效解决传统PD控制器依赖被控对象质量参数的难题。所提出的方法采用自主学习的方式对卫星姿态进行控制,具有很强的智能性和一定的普适性,在未来卫星执行复杂空间任务中的智能控制方面有着很好的应用潜力。     

TopPixelLoss:类别不均衡的遥感影像语义分割损失函数

针对遥感影像中类别不均衡的小目标分割效果不理想的问题，提出了一种类别不均衡小目标二分类分割的损失函数——TopPixelLoss损失函数。首先计算出每个像素的交叉熵，然后将所有像素的交叉熵按从大到小进行排序，随后确定一个K值作为阈值，筛选出前K个交叉熵最大的像素，最后对于筛选出的K个像素交叉熵取平均，做为损失值。在ISPRS 提供的 Vaihingen 数据集上，使用PSPNet网络与普通交叉熵、FocalLoss、TopPixelLoss三种损失函数分别对车辆进行二分类分割试验。结果表明，不同的K值，使用TopPixelLoss损失函数的平均交并比（MIoU）、F1-score、准确度（ACC）都最高;当K值为5×104时效果最佳，MIoU、F1-score、ACC分别比FocalLoss提高了3.0%、5.0%、0.1%。TopPixelLoss损失函数是一种针对类别不均衡分割非常有效的损失函数     

期望状态序列导向的深空探测器规划修复方法

Aiming at the challenges caused by the persistence, concurrency and energy consumption of probe actions, a plan repair method of deep space probe based on the expected state sequence is proposed. In this method, the expected state sequence is formed of the expected effect of the unfinished action and the expected precondition of the unexecuted action in the pre designed plan, according to the execution status of the action. The expected state sequence is an ordered set of states with mixed logic and energy, providing subgoals for plan repair and also transforming the plan repair problem into the state transition path searching problem. During the search, the plan repair strategy with energy supply priority is proposed, which separates the logic repair from energy repair to reduce the difficulty of solving the problem. And this method enables the probe to recover from plan failure autonomously. Finally, the effectiveness and rationality of the proposed method are verified through simulation by taking the Mars Orbiter as an example.     

基于时标分解的弹性高超声速飞行器智能控制

考虑弹性高超声速飞行器纵向动力学模型，提出了一种基于时标分解的智能控制方法。考虑刚体状态和弹性模态具有不同的时标特性，采用奇异摄动理论进行快慢时标分解，将模型转换为刚体慢变子系统和弹性快变子系统。针对刚体子系统考虑动力学不确定，基于平行估计模型构造表征不确定逼近效果的预测误差，结合跟踪误差给出复合学习控制策略。针对弹性子系统设计自适应滑模控制稳定弹性模态。通过李雅普诺夫稳定性分析可证系统状态一致终值有界。仿真表明所提出的控制方法能够实现刚弹模态的稳定收敛，且具有更高的跟踪精度、更好的学习性能和更快的收敛速度。     

Propulsionless planar phasing of multiple satellites using deep reinforcement learning

This work creates a framework for solving highly non-linear satellite formation control problems by using model-free policy optimisation deep reinforcement learning (DRL) methods. This work considers, believed to be for the first time, DRL methods, such as advantage actor-critic method (A2C) and proximal policy optimisation (PPO), to solve the example satellite formation problem of propellantless planar phasing of multiple satellites. Three degree-of-freedom simulations, including a novel surrogate propagation model, are used to train the deep reinforcement learning agents. During training, the agents actuated their motion through cross-sectional area changes which altered the environmental accelerations acting on them. The DRL framework designed in this work successfully coordinated three spacecraft to achieve a propellantless planar phasing manoeuvre. This work has created a DRL framework that can be used to solve complex satellite formation flying problems, such as planar phasing of multiple satellites and in doing so provides key insights into achieving optimal and robust formation control using reinforcement learning.