【摘 要】
：
Vibration reduction has always been one of hot and important topics in mechanical engineering,especially for the special measurement instrument.In this paper,a novel limb-inspired bionic structure is proposed to generate negative stiffness and design a ne
【机 构】
：
State Key Laboratory of Advanced Design and Manufacture for Vehicle Body,Hunan University,Changsha 4
论文部分内容阅读
Vibration reduction has always been one of hot and important topics in mechanical engineering,especially for the special measurement instrument.In this paper,a novel limb-inspired bionic structure is proposed to generate negative stiffness and design a new quasi-zero stiffness isolator via torsion springs,distinguishing from the existing tension spring structures in the literature.The nonlinear mathematical model of the proposed structure is developed and the corresponding dynamic properties are further investigated by using the Harmonic Balance method and ADAMS verification.To evaluate the vibration isolation performance,typical three-springs quasi-zero stiffness (TS QZS) system is selected to compare with the proposed bionic structure.And the graphical processing unit (GPU) parallel technology is applied to perform necessary two-parameter analyses,providing more insights into the effects of parameters on the transmissibility.It is shown that the proposed structure can show advantages over the typical TS QZS system in a wider vibration isolation range for harmonic excitation case and shorter decay time for the impact excitation case.
其他文献
本文为减少三支决策中主观性因素对决策结果的影响,并克服经典多粒度粗糙集模型在信息融合方面存在极端性的局限,在对偶犹豫模糊信息系统中探索了基于可调多粒度概率粗糙集的三支决策模型与方法,用于求解多属性群决策问题.首先,本文将对偶犹豫模糊的概念引入三支决策中,提出了可调多粒度对偶犹豫模糊概率粗糙集模型.然后,本文依据离差最大化法计算属性权重与专家权重,进一步利用多粒度三支决策框架建立了对偶犹豫模糊多属性群决策方法.最后,通过医学诊断的实例验证了本文所建立方法的可行性与有效性.
传统K-means聚类算法初始聚类中心以及聚类数目K是随机确定的,聚类结果受其影响较大,这样容易造成聚类结果不稳定且准确率较低.针对上述问题,本文提出一种基于优化初始聚类中心和轮廓系数的K-means聚类算法.首先,为了选出准确的初始聚类中心,引入平均样本距离和误差平方和,构造初始聚类中心的选取方法,使得选取的初始聚类中心是样本相对集中的点,有效避免选择离群点;然后,为了选择出最佳聚类数目K,基于最近簇中心进行簇的合并,基于中位数构造轮廓系数,设计基于中位数的平均轮廓系数评价指标,判断簇合并之后的最佳K;
借鉴局部粗糙集的理论,本文利用模糊集上包含度的定义提出了局部模糊粗糙集模型,分析了该模型的性质并且给出了该模型的属性约简,最后通过实例说明该属性约简算法是有效的.
邻域粗糙集可以同时处理名义与数值属性,多粒度粗糙集提供多个粒度视角下的 目标概念近似,变精度粗糙集使得近似集计算不再局限于完全包含.本文首先提出了 一种同时具有以上三种粗糙集模型长处并且粒度可变的变精度多粒度邻域粗糙集模型,并设计基于矩阵的近似集计算与更新方法:首先提出静态计算近似集的矩阵算法,继而考虑在邻域粒变小时,基于静态计算算法对近似集进行更新,提出一种邻域粒变小时近似集更新的矩阵算法,最后通过UCI公开数据集实验验证了计算与更新算法的有效性.
多示例学习(MIL)的任务是训练一个有效的分类器,以处理具有复杂数据结构的包.一个包对应一个样本,由多个实例构成,描述了样本的信息特征.基于标准MIL假设,如果包中至少有一个正实例,则该包为正,反之为负.已有的多示例学习算法通常将包看作一个整体或基于整个实例空间进行学习.然而,数据集中通常包含噪声,将对分类结果造成一定的影响.本文提出半监督多示例分类的两层粒化与空间转换方法(TSSM).首先,在单包粒度层次上,设计基于密度与距离的去噪技术,获得特征值更为突出的包.其次,在数据集粒度层次上,设计关键包选择技
本文证明了环上的L-模糊理想和环上的L-模糊环同余关系是一一对应的,模上的L-模糊子模和模上的L-模糊模同余关系是一一对应的.
本文研究q-阶正交模糊环境中广义混合平均算子的多属性决策问题.首先,针对多属性决策时需掌控变量间的权重关系以及减少极端数值对决策结果造成影响的两种需求,本文将q-阶正交模糊数与广义混合平均算子相结合,提出广义q-阶正交模糊混合平均算子.其次,对广义q-阶正交模糊混合平均算子的相关性质进行证明.最后,给出一种基于该算子的多属性决策方法,并用实例验证该方法的可行性和有效性.
The non-ordinary state-based peridynamic (NOSB PD) model has the capability of incorporating existing constitutive rela-tionships in the classical continuum mechanics.In the present work,we first develop an NOSB PD model corresponding to the Johnson-Holmq
In this study,we conduct three-dimensional nonlinear large-eddy simulation to investigate the interaction between turbulence and reaction during the initial ignition process of a turbulent methane/hydrogen jet-in-hot-coflow flame under moderate or intense
We present an experimental study of flow structures in turbulent Rayleigh-Bénard convection in annular cells of aspect ratios Γ=1,2 and 4,and radius ratio ? 0.5.The convecting fluid is water with Prandtl number Pr=4.3 and 5.3.Rayleigh number Ra ranges 4.8