边缘计算中基于博弈论的数据协作缓存策略研究

针对边缘计算中服务器存储能力有限的问题,提出一种基于博弈论的数据协作缓存策略。该策略根据基站覆盖范围将边缘计算环境划分为多个区域,每个区域与相邻区域协作缓存数据资源。在每个区域中,计算每个数据块对本地区域及相邻区域的缓存价值,根据待缓存资源的缓存价值进行缓存决策,最小化用户获取数据资源的延迟。仿真实验结果表明,提出缓存策略比现有非协作缓存策略数据资源平均获取延迟降低了36.55%,有效降低了数据资源平均获取延迟。     

SnAgCu无铅焊点低周疲劳行为研究

在25℃下利用单轴微力疲劳试验机对96.5Sn-3Ag-0.5Cu无铅焊点进行不同频率（1Hz~10Hz）和应变范围（2%~8%）的低周疲劳试验。结果表明,不同应变范围条件下无铅焊点的低周疲劳行为符合Coffin-Manson方程。频率修正的Coffin-Manson方程可以用来描述频率对无铅焊点低周疲劳寿命的影响。疲劳裂纹首先在焊点边缘的钎料与金属间化合物（IMC）之间的界面处萌生,随后,裂纹沿近IMC层的钎料内进行扩展。不同频率条件下焊点的断口形貌主要分为三个特征区域：裂纹萌生区、裂纹扩展区和最终断裂区。随着频率的升高,焊点的断裂机制由沿晶断裂向穿晶断裂转变。     

A new Gd-promoted nickel catalyst for methane conversion to syngas and as an anode functional layer in a solid oxide fuel cell

The effect of lanthanide promoters on a Ni-Al₂O₃ catalyst for methane partial oxidation, steam reforming and CO₂ reforming at 600-850 °C is systematically investigated. The promoters include La₂O₃, CeO₂, Pr₂O₃, Sm₂O₃ and Gd₂O₃. GdNi-Al₂O₃ shows comparable catalytic activity to LaNi-Al₂O₃ and PrNi-Al₂O₃ but higher activity than CeNi-Al₂O₃ and SmNi-Al₂O₃ for all three reactions. The O₂-TPO results show that GdNi-Al₂O₃ possesses the best coke resistance among those tested. It also displays good stability at 850 °C for 300 h. Raman spectroscopy indicates that the addition of lanthanide promoters can reduce the degree of graphitization of the carbon deposited on Ni-Al₂O₃. The GdNi-Al₂O₃ is further applied as an anode functional layer in solid-oxide fuel cells operating on methane. The cell yields peak power densities of 1068, 996 and 986 mW cm⁻² at 850 °C, respectively, for operating on methane-O₂, methane-H₂O and methane-CO₂ gas mixtures, which is comparable to operating on hydrogen fuel. GdNi-Al₂O₃ is promising as a highly coking-resistant catalyst layer for solid-oxide fuel cells.     

Well-defined polypyrrole nanoflakes via chemical oxidative polymerization in the presence of sodium alkane sulfonate

The conductive polymers with unique nanostructures have attracted intense interest due to their potential application. Here the well-defined polypyrrole nanoflakes (PPy/AS) were fabricated via the facile chemical oxidative polymerization of pyrrole with high feeding ratio of sodium alkane sulfonate (SAS). SAS was used as the surfactant (template) and dopant for the nanostructured conductive polymer for the first time. The morphology of the products changed from particles to nanoflakes with the increasing of the feeding ratio of SAS. The polypyrrole nanoflakes (PPy/AS) were characterized with Fourier transform infrared (FTIR), thermogravimetric analysis (TGA), transmission electron microscopy (TEM) and electrical conductivity techniques. They exhibited the weak electrical conductivity dependence on temperature.     

Key meat flavour compounds formation mechanism in a glutathione-xylose Maillard reaction

The formation mechanism of meat flavours formed from a glutathione-xylose Maillard reaction was studied using a group of model reactions with [¹³C₅] xylose/xylose (1:1), heated at 132 °C for 90 min. Volatiles, especially the aroma-active compounds, and non-volatiles were analysed respectively with GC-O-MS and LC-TOF-MS. Analysis of the mass spectra showed that furfural, 2-furfurylthiol and thiophene were ¹³C₅-labelled and hence stem from xylose, whereas 2-methyl-3-furanthiol, 2-methyl-thiophene and 2-pentyl-thiophene, which showed similar formation patterns with only unlabelled compounds, may be from thiamine and/or cysteine carbons. In the process of Maillard reactions, GSH can be capable of cleaving in the position of glutamyl and cysteinyl, and then form 5-oxoproline or pyroglutamic acid (PCA), cysteine, and Cys-Gly dipeptide, which can form cyclic (Cys-Gly).     

Study on proton-conducting solid oxide fuel cells with a conventional nickel cermet anode operating on dimethyl ether

This study investigates dimethyl ether (DME) as a potential fuel for proton-conducting SOFCs with a conventional nickel cermet anode and a BaZr_0.4Ce_0.4Y_0.2O_3−δ (BZCY4) electrolyte. A catalytic test demonstrates that the sintered Ni + BZCY4 anode has an acceptable catalytic activity for the decomposition and steam reforming of DME with CO, CH₄ and CO₂ as the only gaseous carbon-containing products. An O₂-TPO analysis demonstrates the presence of a large amount of coke formation over the anode catalyst when operating on pure DME, which is effectively suppressed by introducing steam into the fuel gas. The selectivity towards CH₄ is also obviously reduced. Peak power densities of 252, 280 and 374 mW cm⁻² are achieved for the cells operating on pure DME, a DME + H₂O gas mixture (1:3) and hydrogen at 700 °C, respectively. After the test, the cell operating on pure DME is seriously cracked whereas the cell operating on DME + H₂O maintains its original integrity. A lower power output is obtained for the cell operating on DME + H₂O than on H₂ at low temperature, which is mainly due to the increased electrode polarization resistance. The selection of a better proton-conducting phase in the anode is critical to further increase the cell power output.     

考虑竞争环境的联邦学习多维契约激励机制研究

针对现有激励机制无法满足竞争环境下联邦学习(FL)落地实施需要的现实问题,提出了一种适于竞争环境的多维契约激励机制。首先,引入了竞争强度描述数据拥有者和任务发布者之间的竞争关系;其次,创新性地设计了一种金钱和FL模型使用权相组合的MM(monetary-the FL model)组合激励;再次,在MM组合激励的基础上,分别在完全信息场景和不完全信息场景下,以任务发布者利润最大化为目标,构建了适于竞争环境的联邦学习多维契约激励机制设计优化问题;进一步,通过理论推导分析了契约可行性及契约最优性,并在此基础上给出了适于竞争环境的联邦学习最优契约设计算法;最后,实验结果表明,以MM组合激励为基础的多维契约激励机制提高了竞争环境下数据拥有者参与联邦学习的意愿。     

基于模拟退火的空域扇区优化方法研究

为了搜索空域扇区优化中的满意解,结合计算几何和模拟退火算法对空域扇区优化问题进行了研究。根据管制空域结构和交通流量空间分布,建立空域扇区分割的模糊多目标函数和约束条件函数,提出划设空域的二分策略,并结合模拟退火算法对扇区优化划设问题进行求解。实例分析表明,结合二分策略的模拟退火方法可获得满意解,扇区划设多目标优化的总体满意度比仅考虑均衡扇区平均流量时提高了2.1%。     

Ti-doped Gd2O3 sensing membrane for electrolyte-insulator-semiconductor pH sensor

The paper reports Ti was added to Gd₂O₃ as pH sensing membrane on silicon combined with proper rapid thermal annealing for the electrolyte-insulator- semiconductor application. It can be found that the high-k Gd₂TiO₅ sensing membrane annealed at 800 °C could obtain high sensitivity, high linearity, low hysteresis voltage, and low drift rate due to improvements of crystalline structures. The high-k Gd₂TiO₅ sensing membrane shows great promise for future bio-medical device applications.     

Physically mixed LiLaNi-Al2O3 and copper as conductive anode catalysts in a solid oxide fuel cell for methane internal reforming and partial oxidation

Different concentrations of copper are added to LiLaNi-Al₂O₃ to improve the electronic conductivity property for application as the materials of the anode catalyst layer for solid oxide fuel cells operating on methane. Their catalytic activity for the methane partial oxidation, steam and CO₂ reforming reactions at 600-850 °C is systematically investigated. Among the three catalysts, the LiLaNi-Al₂O₃/Cu (50:50, by weight) catalyst presents the best catalytic activity. Thus, the catalytic stability, carbon deposition and surface conductivity of the LiLaNi-Al₂O₃/Cu catalyst are further studied in detail. O₂-TPO results indicate that the coking resistance of LiLaNi-Al₂O₃/Cu is satisfactory and comparable to that of LiLaNi-Al₂O₃. The surface conductivity tests demonstrate it is extremely improved for LiLaNi-Al₂O₃ catalyst due to the addition of 50 wt.% copper. A cell with LiLaNi-Al₂O₃/Cu (50:50) catalyst layer is operated on mixtures of methane-O₂, methane-H₂O and methane-CO₂, and peak power densities of 1081, 1036 and 988 mW cm⁻² are obtained at 850 °C, respectively, comparable to the cell with LiLaNi-Al₂O₃ catalyst layer. In summary, the results of the present study indicate that LiLaNi-Al₂O₃/Cu (50:50) catalysts are highly coking resistant and conductive catalyst layers for solid oxide fuel cells.