Ar/CO/NH3等离子体刻蚀多种磁性金属叠层

基于磁性合金的等离子体刻蚀工艺,在中国科学院微电子研究所自主研发的刻蚀设备中,使用Ar,CO和NH3的混合气体对用于形成磁随机存储器(MRAM)的多种磁性金属叠层进行刻蚀.采用两步刻蚀的方法刻蚀了多层磁性金属叠层,每一步的刻蚀气体组分不同,研究了Ar气在混合气体中的体积分数对材料刻蚀速率和侧壁形貌的影响.结果表明,刻蚀速率随着Ar气体积分数的增加而增加,而侧壁倾角则随着Ar气体积分数的增加而减小.两步刻蚀的方法可以根据材料的结构特点来控制Ar离子轰击的作用,可以得到大于21.4 nm/min的刻蚀速率和约90°的侧壁倾角.     

砂岩多步压剪的声发射特性研究

为了了解砂岩多阶段性应力分布特征,进行了多步压剪实验,加载速率分别为0.02 kN/s、0.1 kN/s、0.5 kN/s,并同步采集全过程声发射信号。利用最大似然估计法分析了声发射能量统计分布特征,结果表明:(1)多步压剪试验的能量概率密度函数符合幂定律分布,并满足能量上的幂律无尺度分布;(2)加载阶段与恒压阶段声发射信号量相差较大,跨越能量数量级有所不同,但加载、恒压阶段的概率密度函数与整体概率密度函数具有统一性,并可以用同一个幂值来表征;(3)加载阶段与恒压阶段幂值分布具有相似性,与概率密度函数形成良好对应;(4)恒压状态下的岩石具有流变特征,声发射能量信号在小能量区间服从线性分布;(5)加载后期,由于裂纹的扩展以及成核过程岩石会发生相变,此阶段能量概率密度函数与全过程保持一致,不受之前能量信号影响。     

加权多步波长移相算法研究

波长移相干涉技术通过改变光源的波长实现移相,可以克服硬件移相引入的误差,同时原理上可以实现多组干涉信号的分离。主要应用于大尺寸干涉仪及需要分离多表面干涉信号的系统中。加权多步波长移相算法是其主要的算法,中心思想是通过时域加入合适的窗函数,抑制寄生信号,提出包含轮廓或厚度信息的有用信号。文中介绍其设计思想,进行算法设计,并对一光学件进行了测试和比对。     

在误差分离技术中反向法的测量精确度

精密工件圆度误差和仪器主轴径向旋转误差的分离技术是近十多年来发展起来的新的测量技术,用于提高两者的测量精确度。反向法误差分离技术是其中的一种。本文对反向法的测量精确度做了详细的理论分析和实验验证。从分析和实验验证中我们得出:许多精密工件和主轴系统的圆度误差主要由偶次谐波组成,而反向法是不能适当地分离出偶次谐波的。即除了第一种反向法外,其余的反向法不能完全分离开两者的圆度误差,对于同一工件反向法对多步法的相对测量误差大约为40%,这种测量精确度是不足的。本文的结论是对反向法误差分离技术的测量精确度的一种完全新颖的见解。     

User Independent Paging Scheme for Mobile IP

Multi-step paging has been widely proposed in personal communications services (PCS) systems to reduce the signaling overheads. Similar ideas can be applied to Mobile IP to provide IP paging services. However, current proposed multi-step paging schemes are user dependent under which the partition of paging areas and the selection of paging sequence are different for each user. The performance of a user dependent paging scheme for individual users may be affected by many factors. It is often difficult to achieve perfect performance for each user. In addition, when multiple users are paged at the same time, user dependent paging schemes may consume significant system resources. This paper introduces a user independent paging scheme where the paging criterion is not based on individual user information. The goal of user independent paging is to provide satisfactory overall performance of the whole system, when personalized optimal performance for each user is hard to obtain. The user independent paging scheme is proposed for IP mobility for its easy implementation and convenient combination with paging request aggregation. The paging criterion adopted is the mobility rate of each subnet determined by the aggregated movements of all mobile users. In order to implement the proposed scheme, a concept of “semi-idle state” is introduced and the detailed solution for obtaining mobility rate is presented. Analytical results show that when paging one user at a time, the performance of the proposed user independent paging scheme is comparable to that of the paging schemes based on perfect knowledge of user movement statistics. When paging multiple users simultaneously and when the knowledge on individual user behavior is not perfectly accurate, the proposed scheme has remarkable advantages in terms of reducing the overall paging cost. Jiang Xie received her B.E. degree from Tsinghua University, Beijing, China, in 1997, M.Phil. degree from Hong Kong University of Science and Technology in 1999, and M.S. and Ph.D. degrees from Georgia Institute of Technology in 2002 and 2004, respectively, all in electrical engineering. She is currently an assistant professor with the Department of Electrical and Computer Engineering at the University of North Carolina-Charlotte. Her current research interests include resource and mobility management of wireless networks,QoS provisioning, and next-generation Internet. She is a member of IEEE and ACM.     

混沌时间序列局域偏最小二乘回归多步预测模型

针对混沌序列局域一阶多步预测问题,提出了基于偏最小二乘回归的混沌时间序列局域直接多步预测模型,偏最小二乘用于混沌时序重构相空间中演化轨迹前后相点信息间的建模。该模型克服了以往一阶局域单步预测模型进行多步预测时存在的误差累积,而且能抑制重构相空间中多重共线性的影响,提高了预测精度。试验中使用交叉验证方法将偏最小二乘的提取成分数。通过对Chen’s混沌序列和Mackey-Glass混沌序列的多步预测试验,验证了该模型在混沌时序预测方面具有很好的效果。     

分数阶Fourier变换极值搜索算法研究

 针对步进式搜索算法在分数阶Fourier变换域二维平面效率、精度低下的问题,通过推导分数阶Fourier域极值函数的一阶导数及其快速算法,将全局寻优效果好的混沌优化法和局部搜索能力强的多步拟牛顿法相结合,提出了分数阶Fourier变换极值混合优化算法.并对混合算法的全局收敛性进行了证明.最后,通过仿真实例,验证了混合算法的全局收敛性及其快速收敛能力.混合优化算法的收敛速度和精度均好于步进式搜索法、混沌优化法及步进式-多步拟牛顿混合法.     

基于组合数据清洗与NL-ConvLSTM模型的多步风电功率预测

针对风电数据在采集与传输过程中会产生大量缺失值和异常值,采用DBSCAN算法和最优组内差分法(OIV)组合筛删异常值,随机森林(RF)算法填补缺失值,提升数据准确性;并建立基于以ConvLSTM为单元的编码-预测(EF)网络的风电多气象输入多步预测模型,为了更好利用气象特征信息,在ConvLSTM模型的输入侧添加具有自注意力机制的非局部(NL)模块增强数据特征表现,从而搭建组合数据清洗方法的NL-ConvLSTM多步风电功率预测模型。实验结果表明,该方法能够进一步提高风电功率多步预测精度和稳定性。     

INVERSE METHOD TO DETERMINE SOIL HYDRAULIC PROPERTIES FROM TRANSIENT OUTFLOW EXPERIMENTS

Transient outflow experiments coupling with an inverse method are promising to derive soil hydraulic information. The water retention curves obtained from one-step and multi-step outflow experiments were compared with those from the pressure cell method. We found that in one-step experiments the increment of pressure would reduce the non-uniqueness of inverse technique and that in multi-step experiment the combination of cumulative outflow with pressure head in the objective function would improve the final estimation and also reduce the non-uniqueness of inverse problem.     

多步接枝稀土掺杂/氰酸酯树脂基复合材料固化性能研究

通过多步接枝制得系列表面有机化的掺杂TiO2粒子(M系列粒子),制备了系列氰酸酯树脂(CE)基复合材料。结果表明,M系列掺杂粒子的引入,在降低复合材料活化能,有利于复合材料固化成型的同时,未改变复合材料的固化机理。定量M系列粒子(质量分数≤3%)的引入,优化了复合材料的介电性能,使复合材料介电常数增大66.2%,介电损耗减小50%。