基于改进的卷积神经网络的步态识别

为了提高惯性传感器采集到的序列数据中步态识别的准确率，建立了一个激励层改进的卷积神经网络(CNN)模型。针对三轴加速度传感器对运动太过敏感导致步态周期划分不准确的问题，采用加速度传感器与弯曲度传感器组合获取人体运动信息。将CNN模型中激励层的线性整流函数（ReLU）改进为带泄露线性整流函数(Leaky ReLU)，以解决遇到卷积输出数据小于0时神经元被抑制的问题，进而达到提高步态识别准确率的目的。实验结果表明：激励层优化的CNN模型在行走、上下楼和上下坡五种步态模式下识别率达到了95.79%，与未采用弯曲度传感器的改进CNN模型和未进行激励层改进的CNN模型相比，步态识别率有所提高。     

A high-safety active/passive hybrid control approach for compressor surge based on nonlinear model predictive control

Surge active control can expand the stable operating range of the compressor. However, the difficulty of flow measurement, dynamic uncertainty disturbance, actuator delay characteristics, hard constraints of control variable, and system security measures have not been fully considered in the existing active control system, which significantly hinders its engineering application. Therefore, a nonlinear model predictive surge active control method is first presented based on flow estimator designed by using a continuous-time Kalman filter for dealing with the hard constraint of control variable and the impact of actuator delay of compression system with dynamic uncertainty. Then, a high-safety active/surge passive hybrid control strategy is designed, dominated by the surge active control and supplemented by the surge passive control, to ensure the compression system’s safe and stable operation. Lastly, the simulation results suggest that the flow estimator accurately estimates the compressor flow. When considering the delay impact of the actuators and sensors and measurement noise on the system, the proposed method exhibits stronger robustness than the existing methods. The active/surge passive hybrid control strategy can successfully ensure the compression system's safe and stable operation. This paper is of high practical significance for the engineering application of future compressor surge active control technologies.     

超视距空战仿真中的策略识别

针对超视距空战仿真中敌机策略的识别问题,研究了一种基于案例的策略识别方法。该方法通过构建包含假定的对手任务目标、观测数据、策略等内容的案例库,采用相似度计算选择与新的观测相似的案例库子集,并计算策略概率分布来识别对手策略。实验证明,相比与常规的基于案例推理方法,在案例中增加敌机任务目标提高了策略识别准确率,并且在假定敌机目标不正确时,能修复错误假定并进行策略识别,改善了基于案例策略识别方法的性能。     

基于模型定义的复合材料旁路挤压试验压缩夹具参数化设计

针对民用飞机复合材料许用值试验开展中需要设计大量构型类似而繁杂的试验夹具,为提高夹具设计的效率和准确性、促进设计的标准化,本文总结了 ASTM标准中复合材料旁路挤压试验的压缩试验夹具构型设计和制造等相关信息,采用包含基于模型的定义(Model Based Definition,MBD)信息的CATIA二次开发方法,提出...     

固定翼无人机编队的启发分布式模型预测控制

针对无人机编队飞行控制中队形保持、路径规划、重构、防撞等多重功能需求,采用启发型分布式模型预测控制框架,将需求转化为多目标优化问题中的代价函数与相容性约束求解,实现了对无人机编队飞行的一体化控制.同时,采用分阶段式的轨迹规划算法,融合模型预测轨迹规划与多项式轨迹规划,利用多项式轨迹系数实现高效的邻居轨迹信息交互,从而增大了模型预测控制的预测范围,并降低了规划控制过程中的计算量与通信量.最后,通过9架无人机的协同避障编队仿真,证明了该方法的有效性.     

STABC-IR: An air target intention recognition method based on bidirectional gated recurrent unit and conditional random field with space-time attention mechanism

The battlefield environment is changing rapidly, and fast and accurate identification of the tactical intention of enemy targets is an important condition for gaining a decision-making advantage. The current Intention Recognition (IR) method for air targets has shortcomings in temporality, interpretability and back-and-forth dependency of intentions. To address these problems, this paper designs a novel air target intention recognition method named STABC-IR, which is based on Bidirectional Gated Recurrent Unit (BiGRU) and Conditional Random Field (CRF) with Space-Time Attention mechanism (STA). First, the problem of intention recognition of air targets is described and analyzed in detail. Then, a temporal network based on BiGRU is constructed to achieve the temporal requirement. Subsequently, STA is proposed to focus on the key parts of the features and timing information to meet certain interpretability requirements while strengthening the timing requirements. Finally, an intention transformation network based on CRF is proposed to solve the back-and-forth dependency and transformation problem by jointly modeling the tactical intention of the target at each moment. The experimental results show that the recognition accuracy of the jointly trained STABC-IR model can reach 95.7%, which is higher than other latest intention recognition methods. STABC-IR solves the problem of intention transformation for the first time and considers both temporality and interpretability, which is important for improving the tactical intention recognition capability and has reference value for the construction of command and control auxiliary decision-making system.     

Investigation of the planar midlatitude ionospheric trend under different levels of solar activity

A better understanding of the ionosphere through accurate mathematical models is no doubt a crucial element. This study focuses on the challenging problem of building a model representing the complex structure of the midlatitude ionosphere. Previous studies have shown that a regional planar model is suitable in representing the total electron content (TEC) trend in the midlatitude ionosphere in both hemispheres. In this study, the planar trend model for 12 non-overlapping northern hemisphere regions in three groups of geographically near 4 regions is further investigated under different levels of solar activity; low, moderate and high. To that end, the coefficients of the model are estimated in the least squares sense using total electron content values from global ionospheric maps (GIMs) for the years 2009, 2012 and 2014. Subsequently, these coefficients are used to reconstruct estimated TEC maps which are then compared with actual GIM-TEC by investigating their difference in normalized $L_2$ norm squared sense. The regional planar trend model provides a particularly successful representation in the years 2012 and 2014 for which the solar activity level is the dominant factor determining the TEC trend. Under low solar activity conditions of 2009, other factors such as ocean currents, temperature variations and meteorological phenomena are suspected to have a considerable effect in some regions depending on their geographic location and on seasonal trends in those regions. As an example, studies show that under the influence of the Pacific Decadal Oscillation (PDO) and Siberian High (SH), a significant cooling trend between 2004 and 2018 in autumn is observed in Eurasia, which, in conjunction with the low solar activity levels, may be related to the deviations from the actual GIM-TEC in 2009 in these regions. As solar radiation increases, however, such bottom-side forcings are masked in 2012 and 2014 and these deviations are no longer observed.     

基于LOFAR谱图的水下目标识别方法

基于LOFAR谱图的主成分分析法是目前用于对水下目标识别的重要方法之一,文中详细分析了该方法的基本原理和具体应用过程,并对潜艇辐射噪声的实测数据进行了仿真和分类试验,给出了仿真图和试验结果.结果表明,运用该方法对水下目标进行识别的识别率较高.     

基于可重构标定板的激光与视觉联合标定方法

目前,激光雷达与视觉传感器的联合标定方法包含动态在线标定与静态离线标定两大类.动态在线标定对标定的环境有较高的要求,且标定结果不稳定;静态离线标定通常采用标定板,因而对标定板的要求较高,手动选择与线拟合都易引入或放大误差,故提出了一种基于可重构标定板的激光雷达与视觉传感器的联合标定方法.首先,对传统的标定板进行拆分与重...     

一种针对多光谱遥感图像的桥梁快速识别算法

提出了一种针对多光谱图像中桥梁的识别算法。首先,根据水体和背景地物在不同光谱波段的亮度差异,计算多光谱图像的水体指数得到水体增强图,搜索其具有明显双峰的直方图得到最优阈值,实现河流的完整提取；其次,利用桥梁的存在会导致局部水体的光谱异常,沿河流中间线进行潜在桥梁区域的快速提取；再进一步利用桥梁长度以及与河流的空间关系进行鉴别,有效剔除虚警。利用 SPOT4遥感影像进行实验,结果表明本文算法运算量小,对于多个桥梁的识别具有很好的实用性。