基于改进深度强化学习的虚拟网络功能部署优化算法

针对网络功能虚拟化/软件定义网络(NFV/SDN)架构下,网络服务请求动态到达引起的服务功能链(SFC)部署优化问题,该文提出一种基于改进深度强化学习的虚拟网络功能(VNF)部署优化算法.首先,建立了马尔科夫决策过程(MDP)的随机优化模型,完成SFC的在线部署以及资源的动态分配,该模型联合优化SFC部署成本和时延成本,同时受限于SFC的时延以及物理资源约束.其次,在VNF部署和资源分配的过程中,存在状态和动作空间过大,以及状态转移概率未知等问题,该文提出了一种基于深度强化学习的VNF智能部署算法,从而得到近似最优的VNF部署策略和资源分配策略.最后,针对深度强化学习代理通过ε贪婪策略进行动作探索和利用,造成算法收敛速度慢等问题,提出了一种基于值函数差异的动作探索和利用方法,并进一步采用双重经验回放池,解决经验样本利用率低的问题.仿真结果表示,该算法能够加快神经网络收敛速度,并且可以同时优化SFC部署成本和SFC端到端时延.     

星载SAR斜视模式运动目标方位多通道信号重建方法

在星载方位多通道SAR斜视模式下,方位斜视角度和运动目标的速度分别导致回波多普勒频谱发生2次混叠和通道失衡,影响运动目标方位多通道信号重建.针对该问题,该文提出一种适用于多通道斜视模式下的运动目标的重建方法.首先通过方位向去斜预处理消除了斜视导致的2次多普勒混叠,然后通过修正的多通道重建矩阵来解决目标速度导致的通道失衡.此外,该文还研究了通道冗余情况下的杂波抑制能力,分析了估计速度误差带来的残余相位误差,给出了一种星载方位多通道SAR斜视模式下的运动目标速度快速估计搜索方法.最后,通过点目标仿真验证了方法的有效性.     

Extending $$\tau $$ τ -Lop to model MPI blocking primitives on shared memory

The Journal of Supercomputing - MPI communication optimization is essential for high-performance applications. The communication performance models have made some achievements in improving the...     

Application of intelligent real-time image processing in fitness motion detection under internet of things

The purpose is to study the applicability of digital and intelligent real-time Image Processing (IP) in fitness motion detection under the environment of the Internet of Things (IoT). Given the absence of real-time training standards and possible workout injury problems during fitness activities, an intelligent fitness real-time IP system based on Deep Learning (DL) is implemented. Specifically, the keyframes of the real-time images are collected from the fitness monitoring video, and the DL algorithm is introduced to analyze the fitness motions. Afterward, the performance of the proposed system is evaluated through simulation. Subsequently, the Noise Reduction (NR) performance of the proposed algorithm is evaluated from the Peak Signal-to-Noise Ratio (PSNR), which remains above 20 dB for seriously noisy images (with a noise density reaching up to 90%). By comparison, the PSNR of the Standard Median Filter (SMF) and Ranked-order Based Adaptive Median Filter (RAMF) algorithms are not higher than 10 dB. Meanwhile, the proposed algorithm outperforms other DL algorithms by over 2.24% with a detection accuracy of 97.80%; the proposed system can adaptively detect the fitness motion, with a transmission delay no larger than 1 s given a maximum of 750 keyframes. Therefore, the proposed DL-based intelligent fitness real-time IP algorithm has strong robustness, high detection accuracy, and excellent real-time image diagnosis and processing effect, thus providing an experimental reference for sports digitalization and intellectualization.

     

基于EEMD和LeNet-5的脑电信号情绪识别

随着科学技术的进步,人们对情绪这一概念有了全新的认识,从过去认为情绪来源于“心”逐渐发展到了当下普遍认为情绪来源于“脑”。针对脑电信号所具有的诸多特性,首先通过去除心电、肌电噪声,滤波提取脑电信号中的有用波段;再利用集合经验模态分解算法(Ensemble Empirical Mode Decomposition,EEMD)对脑电信号进行特征提取,利用提取特征通过空间插值法绘制脑电地形图;接着利用LeNet-5算法开展具体情绪识别,并建立模型。最终通过不断地改进模型,显著提高了情绪识别准确率,准确率最高可达80.1%。     

RS类纠删码的译码方法

RS(Reed-Solomon)码可以根据应用环境构造出任意容错能力的码字,有很好的灵活性,且使用RS纠删码作为容错方法的存储系统能达到理论最优的存储效率.但是,与异或(exclusive-OR,XOR)类纠删码相比,RS类纠删码译码计算的时间开销过大,这又很大程度上阻碍了它在分布式存储系统中的使用.针对这一问题,提出了一类RS纠删码的译码方法,该方法完全抛弃了当前大多RS类纠删码译码方法中普遍使用的矩阵求逆运算,仅使用计算复杂度更小的加法和乘法,通过构造译码变换矩阵并在此矩阵上执行相应的简单的矩阵变换,能够直接得出失效码元由有效码元组成的线性组合关系,从而降低译码计算复杂度.最后,通过理论证明了该方法的正确性,并且针对每种不同大小的文件,进行3种不同大小文件块的划分,将划分得到的数据块进行实验,实验结果表明:在不同的文件分块大小情况下,该新译码方法较其他方法的译码时间开销更低.     

结合双语义数据增强与目标定位的细粒度图像分类

细粒度图像分类旨在对属于同一基础类别的图像进行更细致的子类划分,其较大的类内差异和较小的类间差异使得提取局部关键特征成为关键所在。提出一种结合双语义数据增强与目标定位的细粒度图像分类算法。为充分提取具有区分度的局部关键特征,在训练阶段基于双线性注意力池化和卷积块注意模块构建注意力学习模块和信息增益模块,分别获取目标局部细节信息和目标重要轮廓这2类不同语义层次的数据,以双语义数据增强的方式提高模型准确率。同时,在测试阶段构建目标定位模块,使模型聚焦于分类目标整体,从而进一步提高分类准确率。实验结果表明,该算法在CUB-200-2011、FGVC Aircraft和Stanford Cars数据集中分别达到89.5%、93.6%和94.7%的分类准确率,较基准网络Inception-V3、双线性注意力池化特征聚合方式以及B-CNN、RA-CNN、MA-CNN等算法具有更好的分类性能。     

基于改进型能量守恒SOC估算法的电动汽车三段式智能充电方式研究

为研究电动汽车安全、快速的智能充电方式,基于传统能量守恒法,考虑电池容量衰减和电池内阻损失对荷电状态(state of charge,SOC)估算的影响,提出改进型能量守恒SOC估算方法。对比分析几种传统充电方式,根据马斯定理确定最佳充电电流,提出以改进型能量守恒SOC估算法得到的SOC值作为判断依据的电动汽车三段式(小电流充电、脉冲充电、恒压充电)智能充电方式,并建立其仿真模型。结果表明:改进型能量守恒SOC估算法得到的SOC值要小于传统能量守恒法,其更加接近真实SOC值;三段式智能充电方式能根据电池组SOC值的变化智能地选择具体充电方式,实现了对电池的安全、快速充电。提出的基于改进型能量守恒SOC估算的三段式智能充电方式对当前电动汽车充电方式的研究提供了一定的参考价值,为智能充电方式研究效能的提升提供了一种可行方法,也为智能充电理论应用于工程实践打下基础。     

Combined effects of high-temperature rolling and ultrafast cooling on the mechanical properties of low-carbon steel

The role of ultrafast cooling (UFC) on the grain refinement of ferrite, the precipitation behavior of cementite particles and the mechanical properties of a mild steel (Q235 grade) was evaluated by applying laminar cooling and UFC and varying the finish cooling temperature ranges during UFC after hot rolling. While UFC refined the ferrite grains, it accumulated the degeneration of pearlite, resulting in complete disappearance of the laminar pearlite at relatively low finish cooling temperatures. The minimum mean size of spheroidized cementite particles reached ~110?nm. Meanwhile, the enhancement of UFC on tensile strengths of mild steels mainly resulted from the grain refinement of ferrite and the precipitation strengthening of cementite particles; however, the contribution varied with the finish cooling temperature of UFC. A modified Ashby–Orowan model was also used for evaluating the yield strength increment of medium plates. This work will provide a theoretical basis for the diversity control of microstructure and for developing stronger and tougher mild steels by introducing UFC technology after high-temperature rolling.     

Laboratory investigation of the properties of epoxy asphalt rubber (EAR)

Crumb rubber is preliminarily mixed with asphalt in wet process to produce ductile and elastic asphalt rubber (AR), which has been extensively used in high performance bituminous mixtures for road pavement. Epoxy asphalt is a thermosetting polymer modified asphalt with excellent performance and has been widely applied on the pavement of steel bridge decks. Epoxy asphalt rubber (EAR) was prepared by mixing AR with the epoxy (EP). The effect of AR concentration on the phase-separated morphology, viscosity, thermal stability and mechanical properties of the neat EP were compared with that of asphalt. Laser scanning confocal microscopy observations revealed that AR particles disperse in the continuous epoxy phase with co-continuous phase-separated structures in EARs with 40 and 50 wt% AR. However, the phase inverts to continuous asphalt structures with dispersed spherical and co-continuous epoxy phase as AR concentration reaches 60 wt%. The addition of AR increases the viscosity of the neat EP. The thermal stability of the neat EP is improved with the incorporation of AR. The presence of AR decreases the tensile strength of the neat EP, while the elongation at break of the neat EP increases with the increase of AR concentration. The viscosity and mechanical properties of EAR with 50 wt% AR completely satisfy the technical requirement of hot-mix epoxy asphalt binder for steel bridge deck pavements.