空袭兵器的模糊识别法

在分析空袭兵器类型区分不足的基础上,运用模糊识别理论,探索空袭兵器的模糊识别方法。提取空袭兵器特征指标,建立标准模型库,构造了隶属函数,构建空袭兵器的模糊识别模型。结合实例,验证该方法简便有效,可为防空部队射击指挥员提供可靠的辅助决策依据。     

基于被动雷达和红外传感器的数据融合

提出基于被动雷达和红外传感器组成的复合探测系统,利用D-S证据理论提高目标识别率,借助神经网络改善目标跟踪算法的性能,同时保持跟踪滤波的计算结构尽可能简单,尤其是减少了数字计算上的复杂性。仿真结果表明了应用新算法的复合探测系统的有效性。     

雷达辐射源识别的算法研究

针对现有雷达辐射源识别算法的不足,探索雷达辐射源数据库识别的新算法。运用属性测度理论建立了雷达辐射源属性识别模型,结合距离算法、灰关联算法和神经网络样本训练思想,设计出基于属性测度模型的雷达辐射源属性识别算法。通过仿真验证,该算法兼具识别准确度高和识别速度快优点,实用性较强。     

防空监视网络传感器的协作管理

针对防空监视网络的传感器管理问题,讨论了基于多Agent技术的协作管理策略。通过对防空监视网络传感器协作行为的分析,建立了传感器的协作行为模型。结合防空监视传感器的功能,设计了传感器Agent的结构模型。依据防空监视网络传感器的使命要求和完成任务能力,提出了防空监视网络传感器Agent的协商处理策略。     

Selling to strategic and loss‐averse consumers: Stocking,procurement, and product design policies

Motivated by the presence of loss‐averse decision making behavior in practice, this article considers a supply chain consisting of a firm and strategic consumers who possess an S‐shaped loss‐averse utility function. In the model, consumers decide the purchase timing and the firm chooses the inventory level. We find that the loss‐averse consumers' strategic purchasing behavior is determined by their perceived gain and loss from strategic purchase delay, and the given rationing risk. Thus, the firm that is cognizant of this property tailors its inventory stocking policy based on the consumers' loss‐averse behavior such as their perceived values of gain and loss, and their sensitivity to them. We also demonstrate that the firm's equilibrium inventory stocking policy reflects both the economic logic of the traditional newsvendor inventory model, and the loss‐averse behavior of consumers. The equilibrium order quantity is significantly different from those derived from models that assume that the consumers are risk neutral and homogeneous in their valuations. We show that the firm that ignores strategic consumer's loss‐aversion behavior tends to keep an unnecessarily high inventory level that leads to excessive leftovers. Our numerical experiments further reveal that in some extreme cases the firm that ignores strategic consumer's loss‐aversion behavior generates almost 92% more leftovers than the firm that possesses consumers’ loss‐aversion information and takes it into account when making managerial decisions. To mitigate the consumer's forward‐looking behavior, we propose the adoption of the practice of agile supply chain management, which possesses the following attributes: (i) procuring inventory after observing real‐time demand information, (ii) enhanced design (which maintains the current production mix but improves the product performance to a higher level), and (iii) customized design (which maintains the current performance level but increases the variety of the current production line to meet consumers’ specific demands). We show that such a practice can induce the consumer to make early purchases by increasing their rationing risk, increasing the product value, or diversifying the product line. © 2015 Wiley Periodicals, Inc. Naval Research Logistics 62: 435–453, 2015     

基于相关性的OFDM盲识别与参数估计方法

针对非协作通信领域OFDM信号识别难的问题,提出一种基于信号循环前缀相关性的盲识别和时域参数估计算法。通过对OFDM信号循环自相关函数进行分析,在时域上求出延时自相关运算函数的波形,利用循环前缀的相关波峰的特性,以区别OFDM信号和单载波信号,实现对OFDM信号的盲识别。在此基础上,通过采用多个OFDM信号构造循环前缀相关性的累加函数所形成的波峰完成对符号和循环前缀长度等参数的估计。     

全CFRP筋混凝土柱低周反复荷载试验研究

考虑轴压比、剪跨比和配箍形式的影响,对6根全CFRP筋混凝土柱进行低周反复荷载试验,通过分析滞回曲线、骨架曲线、位移延性系数和耗能能力等特征,研究CFRP筋混凝土柱的抗震性能。试验结果表明:CFRP筋混凝土柱具有较强承载能力和变形能力;轴压比和剪跨比是影响试件抗震性能的主要因素,轴压比越小,试件的变形和耗能能力越强,刚度退化越平缓,剪跨比越大,试件的变形能力越强;通用屈服弯矩法计算得到的延性系数能够较好反映CFRP筋混凝土柱的延性,轴压比越小、剪跨比越大,试件的延性越好;新型螺旋式矩形箍筋在试件延性及耗能方面表现良好,可以应用于工程实际。     

融合图嵌入的Smooth-PCANet图像识别算法

主成分分析网络（Principal Component Analysis Network,PCANet）是一种简单的深度学习算法,在图像识别领域具有优秀的性能。然而PCANet在构建网络卷积核时只关注了图像的主分量信息,忽视了近邻像素点之间的位置关系。而通常情况下,图像的相邻像素点具有空间强相关性,因此利用图结构保持像素点的位置近邻关系更有利于网络提取有效特征。因此,我们将图嵌入思想融入PCANet,提出一种新的图像识别算法Smooth-PCANet。为了验证Smooth-PCANet算法的有效性,我们在人脸、手写体字符以及图片等不同数据集上构建实验,并将Smooth-PCANet与多种基于深度学习的图像识别算法作了对比。实验结果证明,Smooth-PCANet算法比PCANet获得了更高的识别性能,并且更有效地避免了过拟合,在小样本训练时具有显著优势。     

无人机小样本条件下遮挡和混淆目标识别方法

针对无人机对地目标识别过程中的小样本问题以及目标存在的遮挡和混淆情况,提出了一种融合自注意力机制的小样本目标识别模型。在利用元学习思想获取小样本学习能力的基础上,将自注意力机制学习目标内部各部分之间的上下文依赖关系引入模型,从而增强目标表征能力,以解决遮挡和混淆情况下有效特征不足的难题。为验证模型效果,通过对基准数据集和无人机航拍数据进一步加工,构建了遮挡和混淆目标数据集,设置了不同的遮挡程度和背景混淆率。通过在不同数据集上的验证,并与深度学习模型对比,证明提出的模型具有更高的学习效率和识别正确率。     

Reaction characteristic of PTFE/Al/Cu/Pb composites and application in shaped charge liner

In this paper, the reaction characteristic and its application in shaped charge warhead of a novel reactive material, which introduced copper (Cu) and plumbum (Pb) into traditional polytetrafluoroethylene/aluminum (PTFE/Al), are studied. The thermal analysis and chemical reaction behavior of the PTFE/Al/Cu/Pb mixture are investigated by Differential Scanning Calorimetry (DSC),Thermo-gravimetry (TG), and X-ray Diffraction (XRD) techniques. Then, the shaped charge liners with PTFE/Al/Cu/Pb reactive materials are fabricated, and the X-ray experiments show that they could form reactive jets with excellent performance under the detonation effects of the shaped charge. Based on that, the penetration experiments of shaped charge with PTFE/Al/Cu/Pb reactive liner against steel plates are carried out, and the results demonstrate that the PTFE/Al/Cu/Pb reactive jets could produce a deeper penetration depth compared to the traditional PTFE/Al reactive jets. Meanwhile, the PTFE/Al/Cu/Pb reactive jets also show significant inner-blast effects, leading to dramatically cracking or fragmentation behavior of the penetrated steel plates. This new PTFE/Al/Cu/Pb reactive liner shaped charge presents enhanced penetration behavior for steel targets that incorporates the penetration capability of a high-density and ductility jet, and the chemical energy release of PTFE-matrix reactive materials.