Spatial Reorientation: Effects of Verbal and Spatial Shadowing

ABSTRACT

Spatially disoriented adults flexibly conjoin geometric information (macroscopic shape) and nongeometric information (e.g., the color of a wall) to re-establish their bearings. It has been proposed that non-geometric information is incorporated into a geometric frame of reference through the use of spatial language. Support for this assumption comes from dual-task studies in which participants failed to use non-geometric features when they shadowed a verbal message concurrent to the reorientation task. These studies were performed in small rectangular spaces. In contrast, our study was performed in a large square room. Experiment 1 showed that verbal shadowing did not disrupt non-geometric feature use in this environment. In Experiment 2, we found that a spatial shadowing task that required the encoding of frequently changing spatial directions impaired reorientation behavior. Our study shows that nongeometric information can be used for reorientation without recourse to linguistic processes, and suggests that the use of non-geometric features is dependent upon a spatial coding process.     

Environmental reference systems for large-scale spaces

Human navigation in well-known environments is guided by stored memory representations of spatial information. In three experiments (N = 43) we investigated the role of different spatial reference systems when accessing information about familiar objects at different locations in the city in which the participants lived. Our results indicate that two independent reference systems underly the retrieval of spatial knowledge. Environmental characteristics, e.g., the streets at an intersection, determine which headings are easier to imagine at a given location and lead to differences in accessibility of spatial information (orientation-specific behavior). In addition, access to spatial information depends on the relative direction of a location with respect to the imagined heading, such that information about locations imagined in front of oneself is easier to access than about locations towards the back. This influence of an egocentric reference system was found for environmental knowledge as well as map-based knowledge. In light of these reference system effects, position-dependent models of spatial memory for large-scale environments are discussed. To account for the simultaneous effect of an environmental and an egocentric reference system, we present a 2-level model of spatial memory access.     

Revisiting Perspective-Taking: Can People Maintain Imagined Perspectives?

Abstract

Three experiments were conducted to examine whether people can adopt and maintain imagined perspectives in the absence of target information. The task used entailed providing information about an imagined perspective in advance of target information to examine whether this would facilitate perspective-taking performance and reduce or eliminate alignment effects that are commonly reported in the literature. The three experiments employed different types of spatial environments: an environment learned from navigating a computer screen (Experiment 1), and an immersive environment that was either remote (Experiment 2) or immediate (Experiment 3) at the time of retrieval. Across the three experiments, results showed that information about an imagined perspective can be utilized ahead of target information. Furthermore, they suggested that alignment effects can be reduced as a result of processing information about perspective ahead of target information, but only when reasoning about specific nonimmediate spatial relations (Experiments 1 and 2). Results are discussed in connection with previous findings on spatial updating and the organizational structure of spatial memory.     

一种LSTM模型预测BC值的空间碎片无控再入预报方法

提出一种利用长短周期记忆(LSTM)神经网络模型动态预测无控再入过程中弹道系数(BC)值实现空间碎片高精度再入时刻预报。通过利用空间碎片两行根数(TLE)、简化通用摄动模型(SGP4)与公开的物体陨落时间作为实测数据样本,利用迭代修正BC值方法构建预测模型的训练集,由此构造用于预测BC值的LSTM模型预测BC,再采用高精度轨道外推动力学模型配合预测BC值预报再入时刻,结果表明基于LSTM模型预测BC的空间碎片再入时刻预报方法是可行的,在95%的置信度内,90天以上的再入时刻预报精度小于10%,30天预报精度小于8%。     

Seeing the Forest or the Trees? Shifting Categorical Effects in Map Memory

People use spatial and nonspatial information to structure memory for an environment. Two experiments explored interactions between spatial and social categories on map memory when mediated by retrieval (Experiment 1) and encoding (Experiment 2) demands. Participants studied a map depicting business locations (including proprietors' race). In Experiment 1, participants completed two memory tasks, one globally focused and the other locally focused. The global task compressed, while the local task expanded, within-category similarity. Furthermore, processing styles carried over to the subsequent task. Experiment 2 emphasized either the spatial or social category during encoding, which increased that category's weighting in memory. These results extend the work of Maddox, Rapp, Brion, and Taylor, suggesting that retrieval and encoding demands can shift how these categories affect spatial memory.     

基于LMI方法的导弹变增益状态反馈H∞控制

针对传统变增益控制方法只适用缓变系统,在整个参数空间不能在理论上保证稳定性和性能的缺陷,提出了一种基于LPV模型的状态反馈H∞控制方法,该方法通过在整个参数空间寻找单一Lyapunov函数以保证系统的稳定性,同时,为了保证系统响应的动态特性,控制过程中把闭环极点配置到指定区域内。控制器的解算可转化成一系列的线性矩阵不等式(LMI),易于实现。最后的仿真结果表明,基于该方法设计的控制器具有良好的性能。     

进气道整流罩保形减阻设计研究

提出了一种尾喷管与进气道整流罩保形设计方案,既保持导弹外形特征不变,又与尾喷管内型面实现一体化保形设计。采用CFD方法对尾喷管及整流罩底部内外流场进行了一体化数值模拟,分析了保形设计对进气道整流罩底阻的影响。结果表明,导弹高速飞行时,采用保形设计能减小进气道整流罩的底阻;补燃室压强越高,进气道整流罩底阻越小,从而验证了该设计方案的可行性。     

导弹模糊增益调参鲁棒飞控系统设计

由于导弹飞控系统的高度非线性、时变性和不确定性,如何设计导弹高性能鲁棒控制系统是一项关键技术。利用模糊逻辑很强的知识综合和推理等特点,基于模糊控制和H∞鲁棒控制方法,建立了一种模糊参数化的增益调参飞控系统鲁棒设计方法,并利用线性矩阵不等式(LMI)技术,将鲁棒增益调参控制器设计转化成了相应的LMI凸优化问题,解决了复杂鲁棒控制器的求解问题。并将该方法应用于导弹鲁棒控制系统设计,仿真结果表明了该设计方法有效性。模糊逻辑、鲁棒控制以及线性矩阵不等式技术的有机结合可以设计具有全局稳定性及鲁棒性能要求的高性能导弹飞控系统。     

知识图谱和大语言模型在航天测控问答系统中的融合应用研究

随着自然语言处理技术的发展,智能化检索与问答系统不断发展,为改善传统知识图谱语义解析能力不足和当前通用大语言模型在垂直领域对知识学习不够深入的问题,提出了融合知识图谱的大语言模型方法,进行了两步优化:首先,利用知识图谱在命名实体识别和关系抽取的基础上,构建大模型提示Prompt模板,进行辅助增强生成,利用图谱中存储的数据提供相关来源;其次,利用低阶适应性微调(LoRA)策略冻结大模型原有参数,增加部分网络参数进行微调训练,优化模型在航天测控领域的知识储备与理解。通过两步改进提高了模型整体在语义解析和知识细节上的理解掌握,结合航天测控领域的相关教材、报告和手册等资料,搭建了知识问答系统,取得了较好的效果,说明了该方法具有一定的应用价值。     

火星大气环境下发动机喷流对飞行器气动特性的影响分析

本文采用计算流体力学(CFD)方法模拟火星大气参数和发动机喷流,研究火星大气对飞行器气动特性的影响,针对发动机喷流对火星大气环境下细长体飞行器的气动特性影响开展了研究。计算结果表明:相较地球大气,火星大气环境计算出的轴向力系数更大,压心系数在跨声速段波动更小。为了进行精细化的火星飞行器气动外形设计,需要进一步优化火星大气计算模型,更加真实地模拟火星大气环境带来的影响。火星大气环境下的发动机喷流也会显著改变上升飞行器的全箭气动特性,影响飞行器表面的压力分布和下游流场区域。火星飞行器受到喷流影响,轴向力系数在Ma=5前变小,在Ma=5后变大,法向力系数以及压心系数在跨声速段会突然减小。     

利用混合模型LSTM-DNN进行全球电离层TEC map的中短期预报

近年来,基于深度学习技术的长短期记忆（long short-term memory, LSTM）网络相关预报算法在空间天气的预测方面得到广泛应用,但存在预测误差随时间堆叠的缺陷,因此只能进行有限的短期预测。为解决这一问题,文章将太阳风参数、太阳黑子数、地磁活动水平指数Ap以及磁暴环电流指数Dst作为预报因子加入模型,建立一个基于LSTM和深度神经网络（deep neural networks, DNN）的混合模型来进行全球电离层TEC map的中短期预报。该模型可以明显减小时间递增对预测误差的影响。测试结果表明,相较于单独的LSTM模型,LSTM-DNN混合模型对24 h电离层预报准确率相近,对48 h电离层预报平均相对精度（RA）由79.30%提升到81.18%,对144 h电离层预报平均相对精度由64.97%提升到77.64%。     

基于在线轨迹规划的混合再入制导方法(英文)

针对在线轨迹规划与跟踪,提出了一种混合再入制导方法。该方法将基于航路点的分段轨迹规划与轨迹跟踪制导有效地结合起来。首先给出再入飞行器无量纲运动方程,建立制导坐标系(GCF),并推导了新坐标系下的经纬度表达式。并将再入飞行过程中各种飞行约束条件转换为控制变量约束。为了加快轨迹优化速度,设计了初始再入飞行轨迹和相关航路点,给出了基于航路点信息的分段轨迹在线规划方法。纵向飞行轨迹跟踪采用基于线性二次型调节器(LQR)的方法,横侧向制导采用横向误差走廊的方法进行控制。仿真结果显示,该方法在线轨迹规划平均计算时间小于0.2秒,且具有较高的制导精度。     

基于机器学习模型FY-3D MWRI海面风速反演

风云三号D星(FY-3D)微波成像仪(MWRI)L1级亮温数据可用于全球海面风速反演,本文讨论了在晴空区和云区使用多元线性统计回归模型和机器学习模型反演海面风速的情况,在晴空区将4 d测试集分别放入多元线性统计回归模型,采用随机森林(RF),支持向量回归(SVR),卷积神经网络(CNN)和Stacking融合(SF)模型对海面风速进行反演,最优的均方根误差(RMSE)分别为1.56、1.31、1.24、1.29和1.27 m/s;在云区2 d测试集上的最优RMSE分别为2.12、1.98、1.87、1.89和1.89 m/s。为了进一步验证晴空区海面风速反演的可靠性,选取美国国家浮标数据中心(NDBC)实测的浮标风速对海面反演风速进行验证,CNN反演风速与NDBC实测风速的RMSE为0.74 m/s,决定系数(R²)为0.80;SF反演风速与NDBC实测风速的RMSE为0.85 m/s,R²为0.74。结果证实了通过机器学习模型能够很好地完成FY-3D MWRI亮温反演全球海面风速的任务。