Method for improving classification performance of neural network based on fuzzy input and network inversion

In order to effectively improve the classification performance of neural network, first architecture of fuzzy neural network with fuzzy input was proposed. Next a cost function of fuzzy outputs and non-fuzzy targets was defined. Then a learning algorithm from the cost function for adjusting weights was derived. And then the fuzzy neural network was inversed and fuzzified inversion algorithm was proposed. Finally, computer simulations on real-world pattern classification problems examine the effectives of the proposed approach. The experiment results show that the proposed approach has the merits of high learning efficiency, high classification accuracy and high generalization capability.

Coalition and distribution in fuzzy dynamic supply chain systems

In this paper, two models of coalition and income's distribution in FSCS (fuzzy supply chain systems) are proposed based on the fuzzy set theory and fuzzy cooperative game theory. The fuzzy dynamic coalition choice's recursive equations are constructed in terms of sup-t composition of fuzzy relations, where t is a triangular norm. The existence of the fuzzy relations in FSCS is also proved. On the other hand, the approaches to ascertain the fuzzy coalition through the choice's recursive equations and distribute the fuzzy income in FSCS by the fuzzy Shapley values are also given. These models are discussed in two parts: the fuzzy dynamic coalition choice of different units in FSCS; the fuzzy income's distribution model among different participators in the same coalition. Furthermore, numerical examples are given aiming at illustrating these models., and the results show that these models are feasible and validity in FSCS.

feedback fuzzy-dvs scheduling of control tasks

To consider the energy-aware scheduling problem in computer-controlled systems is necessary to improve the control performance, to use the limited computing resource sufficiently, and to reduce the energy consumption to extend the lifetime of the whole system. In this paper, the scheduling problem of multiple control tasks is discussed based on an adjustable voltage processor. A feedback fuzzy-DVS (dynamic voltage scaling) scheduling architecture is presented by applying technologies of the feedback control and the fuzzy DVS. The simulation results show that, by using the actual utilization as the feedback information to adjust the supply voltage of processor dynamically, the high CPU utilization can be implemented under the precondition of guaranteeing the control performance, whilst the low energy consumption can be achieved as well. The proposed method can be applied to the design in computer-controlled systems based on an adjustable voltage processor.

M_2RSbO_6:R~(3+)复合氧化物的合成组成结构和性能的研究(M = Ba、Sr、Ca, R = La、Gd、Y、Bi, R' = Sm、Eu、Dy、Ho、Er、Tm)

对于稀土与非稀土所组成的二元复合氧化物的研究国外已有较多的报导。但是，对于稀土和锑的复合氧化物只是近年来才开始有些研究工作。含锑与稀土的多元复合氧化物的报导就更少。本文在我们实验室张静筠等人三元复合氧化物研究的基础上，开展Mo—Sb_2O_5—R_2O_3—R'_2O_3—Bi_2O_3多元体系的研究工作，这对于我国丰产元素稀土和锑的应用以及利用Bi~(3+)的激活与敏化将是有益的。本文按Thornton等人的方法合成了Ba_2BiSbO_6，Ba_2GdSbO_6，按EγΦECEHKO等人的方法合成了M_2RSbO_6 (M = Ba、Sr、Ca, R = La Y)。并以M_2RSbO_6为基质，掺Sm~(3+)、Eu~(3+)、Dy~(3+)、Ho~(3+)、Er~(3+)、Tm~(3+)和Bi~(3+)，研究它们的化学组成，晶体结构与发光性能的关系及规律，Bi~(3+)的荧光和敏作用。同时研究了它们的磁学和热学性能。化学组成的分析结果表明，计算的含量与实验测得的含量符合较好，说明化学反应是按化学计量比进行的。通过X-射线粉沫物相分析和晶胞参数的理论计算确定M_2RSbO_6(M = Ba、Sr、R = La、Y、Gd、Bi)复合氧化物是属于立方钙钛太型化合物。空间群为Fm3m，点群为Oh。用计算机计算了Ca_2YSbO_6的晶胞参数并结合荧光光谱分析确定它属于畸变的单斜钙钛矿，空间群为P_(21)。用磁天平测量了样品M_2RSbO_6 (M = Ba、Sr、Ca; R = Gd、Y、Bi)的磁化率。除Ba_2GdSbO_6是顺磁性物质外共余的都是反磁性的物质。按所用原料Sb_2O_5计算的磁化率与测量值符合较好，表明在所研究的M_2RSbO_6化合物中锑是正五价的。用热重热差分析仪测量了样品在反应中的热性能，观察到在化合物形成的过程中所用原料Sb_2O_3大约在520 ℃左右氧化变为Sb_2O_5。除所用原料碳酸盐分解外没有挥发性的物质，这就进一步证明化学组成分析和磁化率测量的结果是正确的。光学测量的结果表明，所有的磷光体随着激活离子浓度的不同其光谱都发生规律性的变化。对于不同Eu~(3+)浓度的Ba_2YSbO_6:Eu~(3+)和Br_2YSbO_6:Eu~(3+), Bi~(3+)体系用254nm激发时均能观察到Eu~(3+)于595nm的尖峰发射。用基质和Bi~(3+)的激发峰325nm激发时，明显地看到敏化剂Bi~(3+)到Eu~(3+)的能量传递，使Eu~(3+)于595nm的发射大大增强，我们认为Bi~(3+)对Eu~(3+)的敏化作用是由于基质和Bi~(3+)的~1S。→ 3P_1的跃迁吸收了激发的能量，然后无辐射弛豫到Eu~(3+)的激发态~5D_0，产生~5D_0 → 7F_1的磁偶极跃迁。对于不同Eu~(3+)浓度的Sr_2YSbO_6:Eu~(3+)和Sr_2YSbO_6:Eu~(3+), Bi~(3+)体系用245nm激发时均能观察到Eu~(3+)于595nm的尖峰发射。用基质和Bi~(3+)的激发峰335nm激发时，观察到基质和Bi~(3+)对Eu~(3+)具有某种能量传递。敏化作用机理与上述的Ba_2YSbO_6:Eu~(3+)和Ba_2YSbO_6:Eu~(3+), Bi~(3+)体系相同。对于不同Eu~(3+)浓度的Ca_2YSbO_6:Eu~(3+)和Ca_2YSbO_6:Eu~(3+), Bi~(3+)体系用396nm激发时，均能观察到Eu~(3+)于613nm很强的尖峰发射。用基质和Bi~(3+)的激发峰313nm激发时，见到Bi~(3+)和基质对Eu~(3+)具有某种能量传递，这种敏化作用主要是由于基质和Bi~(3+)的3P_1 → ~1S_0的400nm的宽带发射和Eu~(3+)的~7F_0 → ~5L_6的396nm的吸收相匹配产生~5L_6→~5D_0→~7F_2的跃迁。通过对激发光谱和荧光光谱的分析给出了Ca_2Y_(0.96)Eu_(0.04)SbO_6的能级图，从实验上可见，Eu~(3+)的发光强烈地依赖于钙钛矿的结构，当Eu~(3+)在空间群为Fm3m 的Ba_2YSbO_6和Sr_2YSbO_6中处于Oh点对称性时，主要是~5D_0 → ~7F_1的磁偶极跃迁。当Eu~(3+)在空间群为P_(21)的单斜钙钛矿中时，主要是~5D_0 → ~7F_2的电偶极跃迁。对于不同掺杂浓度M_2YSbO_6:R~(13+)(M = Ba、Ca; R' = Sm、Dy、Ho、Er、Tm)体系，通过激发和荧光光谱的研究，合理地确定了谱项。发现基质对Sm~(3+)、Dy~(3+)、Ho~(3+)具有敏化作用。对不同Bi~(3+)浓度的Ca_2YSbO_6:Bi~(3+)，由激发和荧光光谱可见Bi~(3+)具有二个激发带，第一激发带位于240nm处相当于~1S_0 → ~1P_1的跃迁，第二激发带位于315nm处相当于~1S_0 → ~3P_1的跃迁。有一个很强的兰紫色发射位于400nm处相当于~3P_1 →~1S_0的跃迁。

Bi系铜酸盐的固体化学性质与超导性能

本论文基于固态化学原理，以Bi系中2223高Tc相的稳定性为中心，通过对该复合氧化物的超导电性的系统研究，从整体上来认识铜酸盐化合物的超导电性的系统研究，从整体上来认识铜酸盐化合物的超导性能。前四章侧重于对Bi(Pb,Sb)系中的一些基本性质和特殊现象的研究，后面章节则主要是对晶体中Cu-O面内、面外元素替代效应的探讨。

Band crossing in isovalent semiconductor alloys with large size mismatch: First-principles calculations of the electronic structure of Bi and N incorporated GaAs

For large size- and chemical-mismatched isovalent semiconductor alloys, such as N and Bi substitution on As sites in GaAs, isovalent defect levels or defect bands are introduced. The evolution of the defect states as a function of the alloy concentration is usually described by the popular phenomenological band anticrossing (BAC) model. Using first-principles band-structure calculations we show that at the impurity limit the N-(Bi)-induced impurity level is above (below) the conduction- (valence-) band edge of GaAs. These trends reverse at high concentration, i.e., the conduction-band edge of GaAs1-xNx becomes an N-derived state and the valence-band edge of GaAs1-xBix becomes a Bi-derived state, as expected from their band characters. We show that this band crossing phenomenon cannot be described by the popular BAC model but can be naturally explained by a simple band broadening picture.

missing data imputation: a fuzzy k-means clustering algorithm over sliding window

Tianjin University of Technology