结构特征向量灵敏度分析的共轭梯度迭代法

在特征向量灵敏度模态法的基础上,将未知模态对特征向量灵敏度的贡献用线性方程组的解来表示,从而提出了不用求Ｋ＾－１的特征向量灵敏度的共轭梯度迭代法。该方法仅需少数几阶模态和少量的迭代次数,便能获得结构高精度特征向量灵敏度,文中给出几种模态法与本文方法的算例比较,显示了该方法的有效性和精确性。     

重根特征向量导数计算的完备模态法

1 引言特征向量导数是结构动力优化分析和控制系统设计过程中很重要的参数;近些年来,在模型修正以及结构参数识别与诊断中也往往需要特征向量导数的信息:另外,特征向量导数还是“灵敏度分析”的一个重要部分.目前计算特征向量导数的部分方法有Fox(含改进Fox 法)、Nelson 法(含改进Nelson 法)、模态法和扰动迭代法.众所周知,求解特征向量导数的“支配方程”(K-λ_iM)(?)_(i,j)=(λ_iM_(?)+λ_(i,j)M-K_(?))(?)_i=g_i (1)的系数阵为亏秩阵;对于单根为亏秩1;对于m 重根就亏秩m.Fox 法和改进Fox 法是利用     

计算特征向量摄动量的混合基展开法

在结构修改和模型校正中，模态展开法是计算特征向量摄动量的常用方法之一，但当高阶模态被截断时，它会带来很大的截断误差。本文利用已知的有限阶模态，构造了Ｎ维欧氏空间的一个新基－混合基，并将特征向量的摄动量在新基上展开来计算特征向量的一、二阶 摄动量。该方法使得不管截模态个数的多少，其精度总与全模态展开法相同，且计算量都远少于全模记展开法；与改进的部分民开法相比，本方法不要求所截留的模态边连续的低阶模态     

计算亏损系统模态灵敏度的逐层递推演算方法

对不具有完全特征向量系的线性振动亏损系统,在由广义模态理论建立的亏损系统广义模态 空间中,给出了亏损系统广义模态灵敏度分析算法,并推导出以逐层递推方式计算广义模态 展开式系数的简便、快速的演算公式. 给出的数值算例证明了此方法的正确性和有效性.     

惯性平台台体结构实验模态参数识别及灵敏度分析

惯性平台台体的动态特性直接决定着惯性仪表的工作精度和可靠性，模态分析是研究机械系统动态特性的主要方法之一。在概述了实验模态分析理论的基础上，建立了某型号平台台体结构的实验模型，对其进行了实验模态分析。通过对实验结果与有限元计算结果比较，验证了有限元结果较为准确；同时针对结构存在的问题，通过灵敏度分析对结构的动力修改提出了改进意见。     

热结构稳态响应的耦合灵敏度分析方法

研究结构稳态热变形和热应力的灵敏度分析方法，给出了直接法和伴随法两种算法。考虑了温度场的耦合作用，在直接法中需要计算温度场对设计变量的导数，在伴随法中需要计算热载荷对温度场的导数。对尺寸和形状两类设计变量的灵敏度分析算例，验证了本文方法的精度。伴随法在应用程序中的实现，为大型结构优化提供了高效率的灵敏度计算方法。     

采用新型损伤指示因子和结构优化方法的单元损伤识别

在一种新型的损伤指示因子的基础上,采用最小二乘法建立损伤识别问题的优化模型,通过模态综合方法计算结构模态的灵敏度,采用半解析法计算指示因子对损伤参数的灵敏度,最终由序列线性规划方法求解优化问题。数值算例验证了本文方法的可行性。     

考虑粘弹性材料频变特性的复合结构频率响应分析

论文采用基于Layerwise离散层理论的有限元板单元建模粘弹性阻尼复合结构,发展了一种考虑粘弹性材料的频变特性的复合结构频率响应计算方法.该方法首先计算结构的质量矩阵和各个频率点下的结构复刚度矩阵,然后求解运动方程的若干阶复特征对并由此计算各阶模态的传递函数,最后将各模态传函线性叠加得到近似的总传函.为了提高计算效率,论文采用了一种高效的数值方法,即只计算若干频率点下的特征向量与特征值,并计算这些点处特征向量关于频率的高阶导数,通过泰勒系数展开逼近和Rayleigh商式,可求得附近若干频率点处的特征向量和特征值,从而避免了在各个频率点下求解大自由度结构特征方程的问题,可以极大地提高计算效率.对一端简支的三层约束阻尼梁算例进行了分析,并与文献中的结果作对比,结果验证了方法的有效性和计算效率.     

对胡海昌的小参数法的补充

专著提出了计算特征向量摄动的一种非模态展开法，本文对这一方法中参数μ或者矩阵μ如何选取的问题进行了研究，分析了μ的数学本质，并指出对于重特征值问题，μ只要是一个适当选定的标量常数既可，无需是个矩阵。     

动态优化中的灵敏度分析研究

本文利用模态分析技术和向量原理,推导出特征值、特征向量和稳态响应幅值随设计变量变化的灵敏度分析式。对于很难求刚度矩阵的结构,推导出直接以柔度矩阵表达的灵敏度公式,文中给出的一个四层框架结构的算例和试验结果,表明了有关灵敏度算式的正确性。     

Numerical simulation of contact problems in vane machinery by a parametric quadratic programming method

Interference fits are widely used for connecting impeller and shaft assembly that are forced together slowly by pressing. The interference fit design ensures stable balance behavior and allows for positive contact between the impeller and shaft assembly throughout the range of operating speeds. In addition to maintaining radial contact, sufficient net radial interface pressure must remain in order to transmit torque when the rotational speed is very high. Therefore, the interference fit between the impeller and the shaft assembly is one of the most important factors influencing the performance of the turbo unit in the design of turbocharger compressor. A suitable fit tolerance needs to be considered in the structural design. A locomotive-type turbocharger compressor with 24 blades under combined centrifugal and interference fit loading is used for the analysis. The finite-element (FE) parametric quadratic programming (PQP) method developed based on the parametric variational principle (PVP) is used for the analysis of the stress distribution in the three-dimensional (3D) contact problem of impeller. The advantages of the parametric programming method compared with conventional approaches are that the penalty factors can be canceled and that solutions can be obtained directly without tedious iterative procedures such as the general incremental iterative method. To save time in the computation, a~multi-substructure technique is adopted for structural modeling. This not only simplifies the calculation, but also provides a convenient service for process computer-aided design (CAD) by means of FE simulation. The effects of the fit tolerance, coefficient of friction and rotational speed (centrifugal force), wall thickness of the shaft sleeve and the contact stress on the interference-fitting surfaces are studied in detail in the numerical computation. It is found that a nonuniform initial amount of interference in the structural design avoids the relative displacement generated and ensures uniformity of the contact stress. To assure quality of press-fitting, the amount of interference between the shaft sleeve and shaft should be strictly controlled to avoid the rapid increase of the contact stress. The numerical results demonstrate the high accuracy and good convergence of the algorithm presented here, which provides an effective approach that achieves more-reliable interference-fitted connections and more-precise assembly accuracy with lower manufacturing cost in the structural design.     

特征向量导数计算的Lanczos方法

本文对特征向量导数计算的Ｌａｎｃｚｏｓ法进行了改进，指出Ｌａｎｃｚｏｓ法对方程的缩阶效率取决于初始向量的选择，介绍了Ｌａｎｃｚｏｓ方法的发展，提出采用频率移位法选择初始Ｌａｎｃｚｏｓ向量。通过理论分析及计算机仿真说明移位Ｌａｎｃｚｏｓ方法可以将较大的方程组缩为一阶方程，在计算精度与效率上大大优于任选向量Ｌａｎｃｚｏｓ法和迭代Ｌａｎｃ－ｚｏｓ法     

随机激励下受热翼面的模态频率特性试验研究

现代高速飞行器结构热模态频率特性试验研究,对这类飞行器设计校核和飞行安全具有重要意义。根据飞行过程中遭受的气动加热特性设计了瞬态热环境模拟系统,同时,根据高温环境的特点对测试中的激励和测量方式进行了重新设计,成功地将普通激振器应用于高温结构模态试验,最终将热环境模拟系统与振动测试系统组合,形成一套考虑瞬态热影响的热模态试验系统,实现了瞬态热环境下结构模态的地面测试。对一个切尖三角翼测量了各个加热区的温度随加热时间的变化,验证了加热温度控制的精确性;在纯随机激励下对测得的激励和振动响应信号采用短时傅里叶变换(Short Time Fourier Transformation,STFT)进行时变模态参数辨识,获得了前四阶模态频率随加热时间的变化,并与结构有限元数值计算结果进行了比较,试验与计算结果吻合得很好,验证了该试验方法对热模态测试问题的有效性和准确性。通过分别对瞬态和稳态热环境下结构模态频率试验和计算结果的分析,探讨了结构瞬态温度场对模态频率影响的机理,揭示了结构内部存在的热应力和材料属性的变化,是决定模态频率随加热时间变化趋势的内在原因。     

薄板振动特征值问题的一种改进积分方程解法

本文根据一种改进的边界元/有限元混合法求解薄板振动固有频率问题,既避开了标准的边界元法所导致的求解非代数特征值方程的困难,亦能够基本上消除通常的边界元/有限元混合法结果精度受区域内部单元划分影响较大的弊端。文中讨论了迭代算法的收敛问题,并用于薄板固有频率分析。数值结果表明,即便是在域内单元很粗疏划分的情况下,本文的方法仍能给出相当满意的结果。     

An iterative approach for nonproportionally damped systems

Modal analysis of nonproportionally damped linear dynamic systems is considered. Dynamic response of such systems can be expressed by a modal series in terms of complex modes. Normally state-space based methods or approximate perturbation methods are necessary for the computation of complex modes. In this paper, an iterative method to calculate complex modes from classical normal modes for general linear systems is proposed. A simple numerical algorithm is developed to implement the iterative method. The new method is illustrated using a numerical example.     

基于贝叶斯信息融合的岸桥有限元模型修正研究

为建立精确的岸桥有限元模型,研究了基于贝叶斯信息融合的模型修正方法．通过方差分析,确定待修正参数,利用中心复合试验设计获取样本点,根据有限元计算结果与实测的结果残差为目标函数获得响应样本．拟合样本点和响应样本值构建二阶多项式响应面模型,并检验响应面模型的精度．基于贝叶斯理论更新融合系数来优化响应面参数,从而获得修正模型．以宁波大榭3号岸桥为工程背景,对比修正后的模态频率和实测频率,最大频率相对误差不超过5%,进而验证了基于贝叶斯信息融合的动力学有限元模型修正方法的有效性．修正后的有限元模型可进一步应用于岸桥的健康监测和安全评估．     

大规模边界元模态分析的高效数值方法

随着大规模快速边界元计算技术的发展,在复杂结构的动态设计、振动与噪声分析中愈来愈多地采用边界元法,因此求解大规模边界元特征值问题、进行复杂结构和声场模态分析,成为工程应用中一个十分重要,但却极具挑战性的课题,目前国际上还没有十分有效的数值方法.本文针对边界元法中典型的非线性特征值问题,提出了一种通用、高效的数值解法,称为基于预解矩阵采样的Rayleigh-Ritz投影法,记为RSRR.首先,通过求解一系列频域边界元问题来构造特征向量搜索空间,进而可以采用Rayleigh-Ritz投影,将原问题转化为一个可以采用现有方法求解的小规模缩减特征值问题;其次,为了降低Rayleigh-Ritz投影过程的计算量,基于解析函数的Cauchy积分公式,构造了边界元系数矩阵的插值近似方法,以及缩减特征值问题系数矩阵的快速计算方法,给出了插值项数的估计策略;最后,将RSRR与声学快速边界元法结合,应用于大规模吸声结构的复模态分析.数值算例表明,RSRR方法能够可靠地求出给定频段内的全部特征值和特征向量,具有计算效率高、精度高、通用等优点.     

Continuum-Based Shape Sensitivity Analysis for 2D Coupled Atomistic/Continuum Simulations Using Bridging Scale Decomposition

In this paper, we propose the first attempt to perform shape sensitivity analysis for two-dimensional coupled atomistic and continuum problems using bridging scale decomposition. Based on a continuum variational formulation of the bridging scale, the sensitivity expressions are derived in a continuum setting using both direct differentiation method and adjoint variable method. To overcome the issue of discontinuity in shape design due to the discrete nature of the molecular dynamics (MD) simulation, we define design velocity fields in a way that the shape of the MD region does not change. Another major challenge is that the discrete finite element (FE) mass matrix in bridging scale is not continuous with respect to shape design variables. To address this issue, we assume an evenly distributed mass density when evaluating the material derivative of the FE mass matrix. In order to support accuracy verification of sensitivity results using overall finite difference method, we use regular-shaped finite elements and only allow shape change in one direction in our example problems, so that design perturbations can be made to the discrete FE mass matrix. However, the sensitivity formulation is sufficiently general to support irregular-shaped finite elements and arbitrary design velocity fields. The sensitivity analysis results, verified using overall finite difference method, reveal the impact of macroscopic shape design changes on microscopic atomistic responses.     

基于FE重构方法的冲击破碎仿真

提出了一种结合FE和SPH的3D冲击破碎问题仿真方法——FE重构法。通过立方体FE单元填充的方式对任意几何体进行离散,并建立粒子一形心重合模型,然后将用LS-dyna软件对粒子模型进行SPH撞击仿真;基于每一个步长的仿真结果,进行FE单元的重构,并结合失效准则进行失效分析,最终得到FE与SPH混合的仿真结果。针对超高速...     

A frequency–time domain finite element model for tidal circulation based on the least-squares harmonic analysis method

An iterative type harmonic finite element model is developed for solving the full non-linear form of the shallow water equations. The scheme iteratively updates time histories of the non-linear terms which are then harmonically decomposed and used as forcing terms for the linear sets of equations which result from the harmonic separation of the shallow water equations. A least-squares harmonic analysis procedure is used to decompose the non-linear forcing terms. This procedure allows for the very efficient separation of extremely closely spaced harmonics, since it is highly selective with respect to the frequencies it considers. In addition tailoring the procedure and using very specific time steps and sampling periods significantly reduces the number of time samplings points required. In conjunction with the iterative nature of our scheme, the least-squares procedure makes the scheme entirely general, allows for the direct assessment of all tidal constituents, including compound tides, and permits the clear cut and complete investigation of their mutual interaction through the non-linearities. In addition this procedure readily computes very-low-frequency or residual type circulations. The FE formulation used shows a very low degree of spurious oscillations while remaining quite simple to implement. This control on nodal oscillations is especially important due to the energy transfer mechanisms involved in this type of iterative scheme. In an example application the effects of the various non-linear overtide and compound tide type interactions are examined. It is demonstrated that not only are compound tides significant relative to the overtides, but they also influence the overtides.