基于ZMP的双足机器人稳定性分析

引入双足机器人研究中一个重要物理量——零力矩点ZMP（zero Moment Point）,研究其计算和测量方法,利用其分析双足机器人在单腿和双腿支撑时的作用范围,最后在二维和三维平面内对机器人的ZMP计算进行探讨,为稳定步态的设计做了理论铺垫。     

利用Liapanov直接方法判定差分方程稳定性

研究差分方程稳定性的强有力的方法是Liapanov直接方法,这一方法的核心思想是针对所研究的方程,构造出其相应的Liapanov函数,利用它来判定方程的稳定性．     

配电网金属氧化物避雷器的选择

文章阐述了配电网金属氧化物避雷器选择的基本内容,给出了选择结果.     

工业酚水稀释的RBF神经网络预测控制

针对酚水稀释过程中酚水流量不便测量的工程情况,提出了一种基于RBF（径向基）神经网络的预测控制方法.在酚水罐每个液位变化周期内使用计算机定时测量液位高度,并将这些高度值作为目标输出,在每个循环结束时训练RBF网络,利用RBF网络的逼近能力来实时预测酚水流量,从而控制清水的配比,实现酚水质量浓度的稳定.该方法可避免测量误差带来的干扰,提高控制精度.     

石油企业财务风险评价神经网络模型的建立

石油企业财务风险评价是石油行业的重要课题.在确立石油企业财务风险评价指标体系的基础上,建立了基于模糊神经网络的石油企业财务风险评价模型.该评价模型分为模糊量化模块和模糊神经网络(FNN)模块,可根据实际问题进行具体调节,具有学习的功能.经过对200多家石油企业的实证分析,结果表明该模型网络预测误差小,是一种可靠、有效的...     

网络融合边界的电信DoS攻击检测*

为了检测网络融合边界的电信拒绝服务攻击,首先针对攻击的类型分别提出了基于呼叫统计或流量信号分析的检测方法,并采用分步骤的策略设计了检测系统,将两种检测算法分别应用于粗检测步骤,再通过精检测步骤确认攻击并找出攻击源.模拟实验结果表明,该方案能够有效地检测出典型的攻击,并能够实时在线检测,具有良好的实用价值,为应对网络融合所带来的新安全问题提供了一种解决方案.     

基于模糊神经网络的大场景人群密度估计方法 *

提出了一种估计大场景下密集人群密度的方法。该方法根据人类视觉的模糊性原理,认为用模糊集来划分人群密度范围比用确定的方法更符合人眼视知觉的认知方式,利用统计的方法确定灰度共生矩阵各指标对于各个密度类别的隶属函数;设计基于误差反向传播训练算法（BP）的模糊神经网络,计算样本模式对于各个密度类别的隶属度,并根据人群密度变化的时域连续性原理对人群密度范围进行合理估计。实验表明该方法提高了估计精度。     

Thermal analysis of montmorillonites modified with quaternary phosphonium and ammonium surfactants

The thermal stability of seven organically modified montmorillonites (‘organoclays’) has been investigated using differential thermal analysis, differential scanning calorimetry, and thermogravimetry in conjunction with X-ray diffractometry. Six organoclays were synthesised by replacing the interlayer inorganic cations, initially present, with quaternary phosphonium and ammonium surfactant cations. The samples modified with tetrabutylphosphonium (TBP), and butyltriphenylphosphonium (BTPP) ions have an appreciably higher thermal stability than the octadecyltrimethylammonium (ODTMA)-modified clays. Thus, in the case of TBP- and BTPP-modified montmorillonites, the onset temperature of decomposition is close to 300 °C. Samples modified with hexadecyltributylphosphonium (HDTBP) ions have a lower onset temperature of decomposition of 225 °C. In comparison, the onset temperature for ODTMA-montmorillonites (obtained at different concentrations of ODTMA-bromide) ranges from 158 to 222 °C, being highest where the concentration of intercalated surfactant is lowest. The onset temperature for a commercial alkylsilane-treated quaternary ammonium-modified organoclay (S-BEN N-400FP) is 207 °C. The basal spacing of the TBP- and BTPP-modified clays is 1.7–1.8 nm, indicating a monolayer arrangement of quaternary phosphonium ions in the interlayer space, while the value of 2.5 nm for HDTBP-montmorillonite indicates a more open structure. The ODTMA-modified samples have basal spacings ranging from 1.9 to 2.1 nm, indicative of a bilayer to pseudo-trilayer arrangement. The exceptionally high basal spacing of 3.4 nm for the S-BEN N-400FP organoclay might be due to interlayer penetration of organosilane hydrolysis products during synthesis. The thermal properties of organoclays are apparently related to the nature of the surfactants and their arrangement in the interlayer space of montmorillonite.     

基于迁移学习的卷积神经网络通道剪枝

卷积神经网络在计算机视觉等多个领域应用广泛,然而其模型参数量众多、计算开销庞大,导致许多边缘设备无法满足其存储与计算资源要求。针对其边缘部署困难,提出使用迁移学习策略改进基于BN层缩放因子通道剪枝方法的稀疏化过程。本文对比不同层级迁移方案对稀疏化效果与通道剪枝选取容限的影响;并基于网络结构搜索观点设计实验,探究其精度保持极限与迭代结构的收敛性。实验结果表明,对比原模型,采用迁移学习的通道剪枝算法,在精度损失不超过0.10的前提下,参数量减少89.1%,模型存储大小压缩89.3%;对比原剪枝方法,将剪枝阈值从0.85提升到0.97,进一步减少参数42.6%。实验证明,引入迁移策略更易实现充分的稀疏化,提高通道剪枝阈值选取容限,实现更高压缩率;并在迭代剪枝的网络结构搜索过程中,提供更高效的搜索起点,利于快速迭代趋近至搜索空间的一个网络结构局部最优解。     

Protein interaction networks: centrality,modularity, dynamics,and applications

In the post-genomic era, proteomics has achieved significant theoretical and practical advances with the development of high-throughput technologies. Especially the rapid accumulation of protein-protein interactions (PPIs) provides a foundation for constructing protein interaction networks (PINs), which can furnish a new perspective for understanding cellular organizations, processes, and functions at network level. In this paper, we present a comprehensive survey on three main characteristics of PINs: centrality, modularity, and dynamics. 1) Different centrality measures, which are used to calculate the importance of proteins, are summarized based on the structural characteristics of PINs or on the basis of its integrated biological information; 2) Different modularity definitions and various clustering algorithms for predicting protein complexes or identifying functional modules are introduced; 3) The dynamics of proteins, PPIs and sub-networks are discussed, respectively. Finally, the main applications of PINs in the complex diseases are reviewed, and the challenges and future research directions are also discussed.