一种基于强化学习的CDN流量调度系统设计方法

针对传统的CDN流量调度系统大多采用启发式方法或规划方法，存在维护成本高，实时性不足等缺点，提出一种基于深度强化学习的CDN流量调度系统设计框架。该框架基于马尔科夫链设计了故障告警网络来触发调度，建立了基于stacking模型的质量评估奖励函数，并在此基础上对流量调度进行定义和建模，构建了基于DQN的深度强化学习模型。最后，通过仿真实验验证了该调度框架的有效性。     

Effects of SiO2 on the preparation and metallurgical properties of acid oxidized pellets

The effects of SiO₂ content on the preparation process and metallurgical properties of acid oxidized pellets, including compressive strength, reduction, and softening–melting behaviors, were systematically investigated.Mineralogical structures, elemental distribution, and pore size distribution were varied to analyze the mechanism of the effects.The results show that with an increase in SiO₂ content from 3.51 wt%to 7.18 wt%, compressive strength decreases from 3150 N/pellet to 2100 N/pellet and reducibility decreases from 76.5% to 71.4%.The microstructure showed that pellets with high SiO₂ content contained more magnetite in the mineralogical structures.Additionally, some liquid phases appeared, which hindered the continuous crystallization of hematite.Also, the softening–melting properties of the pellets clearly deteriorated as the SiO₂ content increased.With increasing SiO₂ content, the temperature range of the softening–melting zone decreased, and the maximum differential pressure and the comprehensive permeability index increased significantly.When acid oxidized pellets are used as the raw materials for blast furnace smelting, it should be combined with high basicity sinters to improve the softening–melting behaviors of the comprehensive charge.     

低氧LiF–NdF3熔盐体系阳极过程的电化学行为研究

The oxidation of oxygen ions and the generation of an anode effect at a low oxygen content of 150 mg/kg were discussed in this paper. Cyclic voltammetry and square-wave voltammetry tests were conducted to explore the anodic processes of LiF–NdF₃ melt after a lengthy period of pre-electrolysis purification at 1000°C (during which the oxygen content reduced from 413 to 150 mg/kg). The oxidation process of oxygen ions was found to have two stages: oxidation product adsorption and CO/CO₂ gas evolution. The adsorption stage was controlled by diffusion, whereas the gas evolution was controlled by the electrochemical reaction. In comparison with oxygen content of 413 mg/kg, the decrease in the amplitude of the current at low oxygen content of 150 mg/kg was much gentler during the forward scanning process when the anode effect occurred. Fluorine-ion oxidation peaks that occurred at about 4.2 V vs. Li/Li+ could be clearly observed in the reverse scanning processes, in which fluorine ions were oxidized and perfluorocarbons were produced, which resulted in an anode effect.     

定位应用型经济数学课程的教学改革探讨

在对经济数学教学现状分析的基础上,为提高学生应用数学知识解决实际问题的能力,将教学与经济管理领域实际问题相联系,视数学应用为教学改革的重点,在教学目标、教材选取、教学方法、教学手段、教学内容及考试制度方面进行了具体的阐述,并就数学软件的正确定位做出合理的阐述。     

医药洁净室的计算流体动力学模拟及实测对比

采用计算流体动力学(CFD)模拟,研究3个医药洁净室的气流流型及颗粒物浓度.采用标准k-ε双方程模型,通过模拟颗粒物在洁净室的分布,分析合理的污染物散发位置及散发比例.对比参考发尘量下的结果与实测数据的差异,根据实测数据确定污染散发量,通过试验不同的单位容积发尘量,计算对应的洁净度,得到与实测数据吻合最好的单位容积发尘量.在单位容积发尘量为1 000pc·(m3·min)-1计算条件下,各房间的模拟结果与实测数据数值比较吻合,且经CFD预测的3个房间的洁净度变化趋势与实测一致.     

食用蓝藻螺旋藻和葛仙米中金属元素的分析

采用硝酸和高氯酸(4∶1)混合酸消化2种可食用蓝藻螺旋藻和葛仙米,原子吸收光谱法检测螺旋藻和葛仙米中的K、Na、Mg、Ca、Zn、Fe、Cu、Mn、Cr、Pb和Cd共11种金属元素含量。结果表明,2种蓝藻中矿质元素含量均较高,特别是Ca含量高达61 905.132mg·kg~(-1);而有害重金属Cr,Po和Cd含量均远低于国家规定标准,表明这2种食用蓝藻均无重金属富集或污染。本结果显示,螺旋藻和葛仙米是极好的天然富钙营养食品,可作为人体矿质元素和微量元素缺乏症的补给品.     

基于热中子探测评价裂缝的示踪剂特性研究

针对裂缝热中子示踪探测方法,从中子双组扩散理论入手结合蒙特卡洛数值模拟方法,分析压裂前后地层热中子分布影响因素,并模拟不同类型示踪剂及含量的热中子裂缝响应规律,筛选最佳示踪剂并给出其在支撑剂中含量的上、下限。结果表明:压裂前后热中子计数变化量主要受裂缝宽度和标记支撑剂中示踪剂含量的影响;钆对裂缝宽度变化反应最灵敏,钆元素是作为标记支撑剂的最佳示踪元素;随支撑剂中Gd 2O 3含量的增加,热中子计数率降低,热中子计数变化量D增加;示踪剂中氧化钆的上、下限值随裂缝宽度呈指数降低,且当裂缝宽度为1.0 cm时,Gd 2O 3的质量分数上限为0.68%,下限为0.03%。     

关于数学物理方程课程教学的研究

针对目前数学物理方程教学中存在的一些问题,结合课程的教学目的与基本特点,提出注重基础知识回顾,精选教学内容,改进教学方法及考核方式等教学改革措施.以激发学生的学习积极性,提高教学质量.     

基于计算思维的大学计算机基础课程教学

大学计算机基础课程是培养大学生综合素质与和创新能力不可或缺的重要环节,而计算思维的培养更是该课程教学的核心。在分析大学计算机基础课程现状和计算思维特征的基础上,从任课教师的角度探讨了以计算思维为导向的教学内容的设计,并取得了较好的实施效果。     

科技成果社会影响力评估的国际经验及启示

 对英国的卓越研究框架（REF）、荷兰的标准评估协议（SEP）和美国的STAR METRICS这3个典型的科技成果社会影响力评估方案进行比较，识别了其评估对象、评估程序和评估目标方面的差异，重点围绕这3个方案的评估内容指标和评估方法进行了深入分析。结合国际经验，认为中国应当将科技成果社会影响力评估置于公立科研机构绩效评价的核心地位；需要在贯彻国家战略导向和体现评估对象、评估目标和区域差异的基础上开发符合中国国情的科技成果社会影响力评估内容和指标体系；要构建大数据评估并使用数据驱动方法进行科技成果的社会影响力评估。