奇异Hamilton系统矩阵的精细积分法

常规位移有限元的结构振动方程是n个二阶常微分方程组.采用一般交分原理推导,将结构振动问题引入Hamiltoil体系,将得到2n个一阶常微分方程组.精细积分法宜于处理一阶方程,应用于线性定常结构动力问题求解,可以得到在数值上逼近精确解的结果.对于非齐次动力方程,当结构具有刚体位移时,系统矩阵将出现奇异.本文借鉴全元选大元高斯-约当法求解线性方程组的经验,提出全元选大元法求奇异矩阵零本征解的方法,该方法可以简便快速地寻求奇异矩阵零本征值对应的子空间.利用Hamiltoil体系已有研究成果及Hamilton系统的共轭辛正交归一关系,迅速将零本征值对应的子空间分离出来,通过投影排除奇异部分,然后用精细积分法求得问题的解.数值算例表明,该方法对Hamilton系统奇异问题,处理方便,计算量小,易于实现,同时保持了精细算法的优点.     

一维离散结构能带结构与表面态的辛分析方法

本文基于辛几何方法推导了一维离散周期结构、半无穷周期结构和含杂质半无穷周期结构的本征方程,力求建立一个完整的辛分析体系。通过辛分析,将一维离散半无限周期结构转化到一个元胞上求解,大大简化了计算量。对于含杂质半无穷周期结构,结合辛分析和W-W算法,给出求解含杂质半无穷周期结构本征值问题的精确、稳定和高效算法。数值算例说明了本文算法的有效性。     

波导本征问题分析的比例边界有限元方法

引入了一种求解波导本征值问题的高效而精确算法-比例边界有限元方法SBFEM (Scaled Boundary Finite Element Method).该方法的一个特点是只需在边界上进行离散,问题降低一维,使计算工作量大大减少;另一特点是所建立的控制方程为二阶常微分方程,可以解析地求解,使计算精度得到了保证.论文利用变分原理并通过比例边界坐标变换,推导了TE波和TM波波导的比例边界有限元频域方程以及波导动剐度方程,同时给出了波导动刚度矩阵的连分式解形式,通过引入辅助变量进一步得出波导特征值方程并求出波导本征值.以矩形、L形波导和叶型加载矩形波导的本征问题分析为例,通过与解析解及其他数值方法比较,结果表明,此方法具有精度高、计算工作量小的优点,而且随着连分式阶数增加收敛速度快.进一步分析了一类角切四脊正方形波导的传输特性.     

大型结构动力响应的状态方程的Krylov精细时程积分法

提出了一种新的精细时程积分法来求解大型动力系统.结合Krylov子空间法、培德级数近似以及一般载荷的维数扩展法,进一步提高精细时程积分法的计算效率.利用维数扩展法避免计算微分方程特解,并可处理任意载荷.对于大型动力系统,通过Krylov子空间的降维分析将问题转化到一个子空间,计算效率得到极大提高.对于迭代次数N的选择作了详细讨论,进一步提高了计算效率.     

?????????????????????Krylov?????????

提出了一种新的精细时程积分法来求解大型动力系统. 结合Krylov子空间法、培德级数 近似以及一般载荷的维数扩展法,进一步提高精细时程积分法的计算效率. 利用维数扩展法 避免计算微分方程特解,并可处理任意载荷. 对于大型动力系统,通过Krylov子空间的降维 分析将问题转化到一个子空间,计算效率得到极大提高. 对于迭代次数$N$ 的选择作了详细讨论,进一步提高了计算效率.     

比例边界等几何分析方法Ⅰ:波导本征问题

提出比例边界等几何方法 (scaled boundary isogeometric analysis, SBIGA), 并用以求解波导本征值问题. 在比例边界等几何坐标变换的基础上, 利用加权余量法将控制偏微分方程进行离散处理, 半弱化为关于边界控制点变量的二阶常微分方程, 即 TE 波或 TM 波波导的比例边界等几何分析的频域方程以及波导动刚度方程, 同时利用连分式求解波导动刚度矩阵. 通过引入辅助变量进一步得出波导本征方程. 该方法只需在求解域的边界上进行等几何离散, 使问题降低一维, 计算工作量大为节约, 并且由于边界的等几何离散, 使得解的精度更高, 进一步节省求解自由度. 以矩形和 L 形波导的本征问题分析为例, 通过与解析解和其他数值方法比较, 结果表明该方法具有精度高、计算工作量小的优点.     

二维非局部线弹性平面问题的辛分析

将二维非局部线弹性理论引入到Hamilton体系下,基于变分原理推导得出了二维线弹性理论的对偶方程和相应的边界条件.在分析验证对偶方程的准确性的基础上,该套方法被应用于二维弹性平面波问题的求解.将精细积分与扩展的W-W算法相结合在Hamilton体系下建立了求解平面Rayleigh波的数值算法.从推导到计算的保辛性确保了辛体系非局部理论与算法的准确性.通过对不同算例的数值计算,分析和对比了非局部理论方法与传统局部理论方法的差别,并进一步指出了该套算法的适用性和优势所在.     

精细积分方法的发展与扩展应用

钟万勰院士于1991年首先提出计算矩阵指数的精细积分方法,其要点是2^N类算法和增量存储。精细积分方法可给出矩阵指数在计算机意义上的精确解,为常微分方程的数值计算提供了高精度、高稳定性的算法,现已成功应用于结构动力响应、随机振动、热传导以及最优控制等众多领域。本文首先介绍矩阵指数精细积分方法的提出、基本思想和发展;然后依次介绍在时不变/时变线性微分方程、非线性微分方程以及大规模问题求解中发展起来的各种精细积分方法,分析了其优缺点和适用范围;最后介绍了精细积分方法的基本思想在两点边值问题、椭圆函数和病态代数方程等问题的扩展应用,进一步展示了该思想的特色。     

Duhamel项的精细积分方法在非线性微分方程数值求解中的应用

基于Duhamel项的精细积分方法,构造了几种求解非线性微分方程的数值算法。首先将非线性微分方程在形式上划分为线性部分和非线性部分,对非线性部分进行多项式近似,利用Duhamel积分矩阵,导出了非线性方程求解的一般格式。然后结合传统的数值积分技术,例如Adams线性多步法等,构造了基于精细积分方法的相应算法。本文算法利用了精细积分方法对线性部分求解高度精确的优点,大大提高了传统算法的数值精度和稳定性,尤其是对于刚性问题。本文构造的算法不需要对线性系统矩阵求逆,可以方便的考察不同的线性系统矩阵对算法性能的影响。数值算例验证了本文算法的有效性,并表明非线性系统的线性化矩阵作为线性部分是比较合理的选择。     

分数阶微分方程的Haar小波算法研究

在BPFs的Caputo分数阶微分算子矩阵的基础上,建立了Haar小波的分数阶微分算子矩阵,提出了一种有效的求解分数阶微分方程的Haar小波数值方法,并将该方法应用于线性和非线性分数阶常微分方程求解中.数值算例表明,该算法简单,数值精确度高,是一种高效的数值求解方法.     

A PRECISE METHOD FOR SOLVING WAVE PROPAGATION IN HOLLOW SANDWICH CYLINDERS WITH PRISMATIC CORES

Wave propagation in infinitely long hollow sandwich cylinders with prismatic cores is analyzed by the extended Wittrick-Williams(W-W) algorithm and the precise integration method(PIM). The effective elastic constants of prismatic cellular materials are obtained by the homogenization method. By applying the variational principle and introducing the dual variables, the canonical equations of Hamiltonian system are constructed. Thereafter, the wave propagation problem is converted to an eigenvalue problem. In numerical examples, the effects of the prismatic cellular topology, the relative density, and the boundary conditions on dispersion relations,respectively, are investigated.     

基于辛本征值方法的受约束介电弹性体中褶皱不稳定性的分析

辛弹性力学已广泛应用于弹性学中各种边值问题的精确解、计算表面波模式以及预测多层超弹性薄膜中的表面褶皱。本文展示了辛分析框架还可应用于受约束介电弹性体中的表面褶皱。机械和电位移向量是两个基本变量来描述介电弹性体中机械变形与电场紧密耦合。褶皱的临界电压可以通过引入基本变量的对偶变量来从辛本征值问题中解决。本文采用扩展的W-W（Wittrick-Williams）算法和精确的积分方法,准确而高效地解决制定的辛本征值问题的本征值。通过将褶皱电压和波数与有无表面能的褶皱基准结果进行比较,验证了辛分析的有效性。辛分析框架简洁且适用于其他不稳定问题,如分层电介质弹性体、磁弹性不稳定性以及层压复合结构的微观和宏观不稳定性。     

多层地基条带基础动力刚度矩阵的精细积分算法

提出应用精细积分算法计算多层地基的动力刚度问题. 精细积分是计算层状介质中波传播的高效而精确的数值方法. 利用傅里叶积分变换将层状地基的波动方程转换为频率-波数域内的两点边值问题的常微分方程组, 运用精细积分方法求解格林函数, 最后再将得到的频率-波数域内地基表面的动力刚度矩阵转换到频率-空间域内, 进而得到刚性条带基础频率域的动力柔度或刚度矩阵. 所建议的精细积分算法, 可以避免一般传递矩阵计算中的指数溢出问题, 对各种情况有广泛的适应性, 计算稳定, 在高频段可以保障收敛性, 并能达到较高的计算精度.      

Coupling of high order multiplication perturbation method and reduction method for variable coefcient singular perturbation problems

Based on the precise integration method(PIM), a coupling technique of the high order multiplication perturbation method(HOMPM) and the reduction method is proposed to solve variable coefcient singularly perturbed two-point boundary value problems(TPBVPs) with one boundary layer. First, the inhomogeneous ordinary diferential equations(ODEs) are transformed into the homogeneous ODEs by variable coefcient dimensional expansion. Then, the whole interval is divided evenly, and the transfer matrix in each sub-interval is worked out through the HOMPM. Finally, a group of algebraic equations are given based on the relationship between the neighboring sub-intervals, which are solved by the reduction method. Numerical results show that the present method is highly efcient.     

Coupling of high order multiplication perturbation method and reduction method for variable coefficient singular perturbation problems

Based on the precise integration method （PIM）, a coupling technique of the high order multiplication perturbation method （HOMPM） and the reduction method is proposed to solve variable coefficient singularly perturbed two-point boundary value prob lems （TPBVPs） with one boundary layer. First, the inhomogeneous ordinary differential equations （ODEs） are transformed into the homogeneous ODEs by variable coefficient dimensional expansion. Then, the whole interval is divided evenly, and the transfer ma trix in each sub-interval is worked out through the HOMPM. Finally, a group of algebraic equations are given based on the relationship between the neighboring sub-intervals, which are solved by the reduction method. Numerical results show that the present method is highly efficient.     

Stationary random waves propagation in 3D viscoelastic stratified solid

Propagation of stationary random waves in viscoelastic stratified transverse isotropic materials is investigated. The solid was considered multi-layered and located above the bedrock, which was assumed to be much stiffer than the soil, and the power spectrum density of the stationary random excitation was given at the bedrock. The governing differential equations are derived in frequency and wave-number domains and only a set of ordinary differential equations ( ODEs) must be solved. The precise integration algorithm of two-point boundary value problem was applied to solve the ODEs. Thereafter, the recently developed pseudo-excitation method for structural random vibration is extended to the solution of the stratified solid responses.     

Random wave propagation in a viscoelastic layered half space

An extremely efficient and accurate solution method is presented for the propagation of stationary random waves in a viscoelastic, transversely isotropic and stratified half space. The efficiency and accuracy are obtained by using the pseudo excitation method (PEM) with the precise integration method (PIM). The solid is multi-layered and located above a semi-infinite space. The excitation sources form a random field which is stationary in the time domain. PEM is used to transform the random wave equation into deterministic equations. In the frequency-wavenumber domain, these equations are ordinary differential equations which can be solved precisely by using PIM. The power spectral densities (PSDs) and the variances of the ground responses can then be computed. The paper presents the full theory and gives results for instructive examples. The comparison between the analytical solutions and the numerical results confirms that the algorithm presented in this paper has exceptionally high precision. In addition, the numerical results presented show that: surface waves are very important for the wave propagation problem discussed; the ground displacement PSDs and variances are significant over bigger regions in the spatial domain when surface waves exist; and as the depth of the source increases the ground displacement PSDs decrease and the regions over which they have significant effect become progressively more restricted to low frequencies while becoming more widely distributed in the spatial domain.     

线性结构的基于市场机制的鲁棒控制

系统建模不可避免地要忽略一些因素从而造成模型误差,因此基于不确定性的非精确模型来设计控制器显得尤为重要.随着各国学者的不断研究,在线性系统H_∞控制方面取得了很大的发展,但仍存在求解过程繁琐、系统的保守性不强、结构复杂和控制器阶数较高等问题.运用计算结构力学与最优控制相模拟的理论来试图完善存在的问题.导引模就是结构力学中的本征值问题,即弹性稳定的欧拉(Euler)临界力或结构振动的本征频率;通过引入区段混合能的概念,采用精细积分和扩展的威廉姆斯(W-W)算法计算导引模,将此方法引入到基于市场机制的控制(market-based control,MBC)理论中,提出了线性结构基于市场机制的鲁棒控制策略.同时给出了在鲁棒控制下,采用李雅普诺夫(Lyapunov)直接法证明了结构基于市场机制的鲁棒控制器的稳定性及参数的确定方法.最后对一高层受控结构进行数值计算与分析,并与H_∞鲁棒算法的控制效果进行了对比.结果表明,导引模不能取到临界值,否则会导致控制输入趋于无穷大.基于市场机制的鲁棒控制效果要好于H_∞鲁棒算法,并且具有较强的应变能力和在线计算时间短等优点,能较好地适用于高层和大跨结构.      

Nonlinear aeroelastic analysis of an airfoil-store system with a freeplay by precise integration method

The aeroelastic system of an airfoil-store configuration with a pitch freeplay is investigated using the precise integration method (PIM). According to the piecewise feature, the system is divided into three linear sub-systems. The sub-systems are separated by switching points related to the freeplay nonlinearity. The PIM is then employed to solve the sub-systems one by one. During the solution procedures, one challenge arises when determining the vibration state passing the switching points. A predictor-corrector algorithm is proposed based on the PIM to tackle this computational obstacle. Compared with exact solutions, the PIM can provide solutions to the precision in the order of magnitude of 10⁻¹². Given the same step length, the PIM results are much more accurate than those of the Runge–Kutta (RK) method. Moreover, the RK method might falsely track limit cycle oscillations (LCOs), bifurcation charts or chaotic attractors; even the step length is chosen much smaller than that for the PIM. Bifurcations and LCOs are obtained and analyzed by the PIM in detail. Interestingly, it is found that multiple LCOs and chaotic attractors can exist simultaneously. With this magnitude of precision and efficiency, the PIM could become a solution technique with excellent potential for piecewise nonlinear aeroelastic systems.