基于特征互补图像的人脸识别仿真研究

研究人脸识别精度问题。由于人脸图像中存在大量干扰信息的缺点,而造成了人脸识别正确率下降,为了解决上述问题,提出了一种基于特征互补图像快速特征融合算法。算法通过对人脸图像的位平面切片图像分析,采用位平面图像分解法,通过各种合成策略构造多幅样本图像。并突出高位平面图像,采用两种加权策略将每一幅人脸图像样本都生成"特征互补图像"。然后,直接用图像的二维典型相关分析(2DCCA)法对两种特征互补图像进行特征抽取。最后通过在ORL国际标准人脸库上进行的实验,结果表明,高位平面图像的典型相关鉴别特征提高了正确识别率,并且因为摒弃了原始人脸图像的大部分干扰信息所以具有更强的鲁棒性。     

Advantages of high-field tokamaks for fusion reactor development

High-field designs could reduce the cost and complexity of tokamak reactors. Moreover, the certainty of achieving required plasma performance could be increased. Strong Ohmic heating could eliminate or significantly decrease auxiliary heating power requirements and high values of n_E could be obtained in modest-size plasmas. Other potential advantages are reactor operation at modest values of , capability of higher power density and wall loading, and possibility of operation with advanced fuel mixtures. Present experimental results and basic scaling relations imply that the parameterB ²a, where B is the magnetic field and a is the minor radius, may be of special importance. A superhigh-field compact ignition experiment with very high values ofB ²a (e.g.,B ²a=150 T² m) has the potential of Ohmically heating to ignition. This short-pulse device would use inertially cooled copper plate magnets. Compact engineering test reactor and/or experimental hybrid reactor designs would use steady-state, water-cooled copper magnets and provide long-pulse operation. Design concepts are also described for demonstration/commercial reactors. These devices could use high-field superconducting magnets with 7–10 T at the plasma axis.     

The need and prospects for improved fusion reactors

Conceptual fusion reactor studies over the past 10–15 yr have projected systems that may be too large, complex, and costly to be of commercial interest. One main direction for improved fusion reactors points toward smaller, higher-power-density approaches. First-order economic issues (i.e., unit direct cost and cost of electricity) are used to support the need for more compact fusion reactors. The results of a number of recent conceptual designs of reversed-field pinch, spheromak, and tokamak fusion reactors are summarized as examples of more compact approaches. While a focus has been placed on increasing the fusion-power-core mass power density beyond the minimum economic threshold of 100–200 kWe/tonne, other means by which the overall attractiveness of fusion as a long-term energy source are also addressed.Nomenclature a Plasma minor radius at outboard equatorial plane (m) - A Plasma aspect ratioR _T /a - AC Annual charges ($/yr) - b Plasma minor radius in vertical direction (m) - B Magentic field at plasma or blanket (T) - B _c Magnetic field at the coil (T) - B Toroidal magnetic field (T) - B Poloidal magnetic field (T) - BOP Balance of plant - C Coil - COE Cost of electricity (mills/kWeh) - CRFPR Compact RFP reactor - CT Compact torus (FRC or spheromak) - c _FPC Unit cost of fusion power core ($/kg) - DC Direct cost ($) - DZP Dense Z-pinch - E Escalation rate (1/yr) - EDC Escalation during construction ($) - ET Elongated tokamak - F Annual fuel charges ($/yr) - FC Component of UDC not strongly dependent or FPC size ($/kWe) - FW First wall - FPC Fusion power core - f _Aux Fraction of gross electric power recirculated to BOP - f ₁ (IC+IDC+EDC)/DC - f ₂ (O&M + SCR + F)/AC - IC Indirect cost ($) - IDC Interest during construction ($) - I _w Neutron first-wall loading (MW/m²) - i Toroidal plasma current (MA) - j Plasma current density, I/a² - k _B Boltzmann constant, 1.602(10)^–16 (J/keV) - LWR Light-water (fission) reactor - MPD Mass power density 1000P_E/M_FPC (kWe/tonne) - M _N Blanket energy multiplication of 14.1-MeV neutron energy - M _FPC Mass of fusion power core (tonne) - n Plasma density (m^–3) or toroidal MHD mode number - O&M Annual operating and maintenance cost ($/yr) - p _f Plant availability factor - PFD Poloidal field dominated (CTs, RFP, DZP) - P Construction time (yr) - P_TH Thermal power (MWt) - P _E Net electric power (1-)P _ET (MWe) - P_ET Total gross electric power (MWe) - p_f Fusion power (MW) - q Tokamak safety factor (B /B _gq )(a/R _T ) - q _e EngineeringQ value, 1/e - R _T Major toroidal radius (m) - RFP Reversed-field pinch - RPE Reactor plant equipment (Account 22) - S Shield - SCR Annual spare component cost ($/yr) - SSR Second stability region for the tokamak - S/T/H Stellarator/torsatron/heliotron - ST Spherical tokamak or spherical torus - T Plasma temperature (keV) - TDC Total direct cost ($) - TOC Total overnight cost ($) - UDC Unit direct cost,TDC/10 ³ P _E ($/kWe) - V _p Plasma volume (m³) - W _p Plasma energy (GJ) - W _B Magnetic field energy (GJ) - Magnetic utilization efficiency, 2nk_BT/(B ²/2₀) - ₀ Permeability of free space, 4(10)^–7 H/m - XE Plasma confinement efficiency, a²/4_E - _e Plasma energy confinement time - _p Overall plant efficiency, _TH(1-) - _TH Thermal conversion efficiency - _FPC AverageFPC mass density (tonne/m³) - Plasma vertical elongation factor,b/a - Thickness of allFPC engineering structure surround plasma (m) - Total recirculating power fraction, (P _ET-P _E)/P _ET, or inverse aspect ratioa/R _T This work was performed under the auspices of USDOE, Office of Fusion Energy.     

Effective DOA Estimation Under Low Signal-to-Noise Ratio Based on Multi-Source Information Meta Fusion

Efficiently performing high-resolution direction of arrival (DOA) estimation under low signal-to-noise ratio (SNR) conditions has always been a challenge task in the literatures. Obvi-ously, in order to address this problem, the key is how to mine or reveal as much DOA related in-formation as possible from the degraded array outputs. However, it is certain that there is no per-fect solution for low SNR DOA estimation designed in the way of winner-takes-all. Therefore, this paper proposes to explore in depth the complementary DOA related information that exists in spa-tial spectrums acquired by different basic DOA estimators. Specifically, these basic spatial spec-trums are employed as the input of multi-source information fusion model. And the multi-source in-formation fusion model is composed of three heterogeneous meta learning machines, namely neural networks (NN), support vector machine (SVM), and random forests (RF). The final meta-spec-trum can be obtained by performing a final decision-making method. Experimental results illus-trate that the proposed information fusion based DOA estimation method can really make full use of the complementary information in the spatial spectrums obtained by different basic DOA estim-ators. Even under low SNR conditions, promising DOA estimation performance can be achieved.     

基于改进MobileNet v2的垃圾图像分类算法

针对现有的垃圾图像分类模型实时性能差和分类精度低的问题,提出基于改进MobileNet v2的垃圾图像分类方法,构建以MobileNet v2为核心的轻量级特征提取网络. 通过调整宽度因子降低模型的参数量;在模型中嵌入通道和空间注意力模块,增强网络对特征的细化能力;设计多尺度特征融合结构,增强网络对尺度的适应性;利用迁移学习的方式优化模型参数,进一步提高模型精度. 实验结果表明,算法在自建数据集上的平均准确率为94.6%,分别高于MobileNet v2、VGG16、GoogleNet、ResNet50、ResNet101模型2.0%、3.4%、3.2%、2.3%、1.2%;所提算法在2种公共图像分类数据集CIFAR-100和tiny-ImageNet中均取得不错表现;模型参数量仅为0.83 M,体积约为基础模型的2/5,在边缘设备JETSON TX2上的单次推理耗时68 ms,实现了推理速度和预测准确率的提升.     

一种基于VINS/FINS组合导航方法

针对视觉惯性组合导航系统中微惯性器件精度偏低,以及足部惯性导航系统航向角误差可观测性差的问题,研究了一种基于上述两种系统的信息双向融合的导航定位方案. 该方法的系统结构由安装于双足步行机器人躯干部分的惯性导航系统和安装于其足部惯性导航系统两部分组成. 惯性导航系统通过视觉同时定位与地图构建数据融合方法可以获得相对准确的航向角,足部惯性导航系统利用零速修正后的位置信息实时修正惯性导航系统中的低精度惯性器件误差,从而构建视觉与惯性信息双向融合的组合导航系统结构. 实验结果表明,该组合导航方案可以有效提高双足步行机器人的航向精度和定位精度.     

基于Rough集的磨削参数决策系统

本文介绍了5层模糊神经网络的优缺点，提出了基于Rough集构造模糊神经网络的方法，并应用于多传感器的磨削参数决策系统，达到控制磨削加工质量的目的。     

基于CCA的人耳和侧面人脸特征融合的身份识别*

鉴于人耳和人脸特殊的生理位置关系,从非打扰识别的角度出发,提出仅采集侧面人脸图像,利用典型相关分析的思想提取人耳和侧面人脸的关联特征,进行人耳和侧面人脸在特征层的融合.实验结果表明,此方法与单一的人耳或侧面人脸特征识别比较,识别率得到提高.     

基于多模型融合的CNN-LSTM-XGBoost短期电力负荷预测方法

短期电力负荷的精准预测可以有效指导机组组合调度、经济调度与电力市场运营。针对输入数据特征量受限时负荷预测的低精度问题,提出一种基于多模型融合的CNN-LSTM-XGBoost短期电力负荷预测方法。通过建立融合局部特征预提取模块的LSTM（long short term memory）网络结构,并将其与XGBoost（eXtreme boosting system）预测模型并行结合,之后结合MAPE-RW（mean absolute percentage error-reciprocal weight）算法进行模型融合初始权重设置,对最佳权重进行搜索,构建最佳融合模型。通过运用电力负荷数据对所提方法进行预测实验,结果表明CNN-LSTM- XGBoost模型的MAPE（mean absolute percentage error）与RMSE（root mean square error）分别为0.377%与148.419 MW,相比于单一网络模型与融合模型结构实现了误差指标的显著降低,验证了基于多模型融合的CNN-LSTM-XGBoost短期电力负荷预测方法具有较快的模型训练速度、较高的预测准确度与较低的预测误差。     

Is LRP2 Involved in Leptin Transport over the Blood-Brain Barrier and Development of Obesity?

The mechanisms underlying the transport of leptin into the brain are still largely unclear. While the leptin receptor has been implicated in the transport process, recent evidence has suggested an additional role of LRP2 (megalin). To evaluate the function of LRP2 for leptin transport across the blood-brain barrier (BBB), we developed a novel leptin-luciferase fusion protein (pLG), which stimulated leptin signaling and was transported in an in vitro BBB model based on porcine endothelial cells. The LRP inhibitor RAP did not affect leptin transport, arguing against a role of LRP2. In line with this, the selective deletion of LRP2 in brain endothelial cells and epithelial cells of the choroid plexus did not influence bodyweight, body composition, food intake, or energy expenditure of mice. These findings suggest that LRP2 at the BBB is not involved in the transport of leptin into the brain, nor in the development of obesity as has previously been described.