群桩作用机理的数值试验研究

 在分析桩土支撑体系及其相互作用关系的基础上，利用有限元分析软件ANSYS建立了群桩 体系计算模型. 通过对均质土体例题计算结果分析，可以看出利用该建模方法与传统解答有 很好的一致性. 根据辽宁工程技术大学实验馆场地土实测资料，利用该建模方法，求解出了承台与桩的 荷载分担比，对安全、经济地进行群桩基础设计具有重要指导作用和应用价值.     

岩土材料应变局部化的有限元分析方法

采用有限单元法分析岩土材料的应变局部化时经常会遇到单元尺寸敏感性问题和网格锁定问题。自适应网格技术能够有效地解决网格锁定问题,但仍然无法克服计算结果对单元尺寸的依赖性,尽管在一维情况下被证明是可行的。复合体理论(均匀化理论)和弱非连续有限元方法可以成功地解决岩土材料的单元尺寸敏感性问题,在一维情况下两类方法实际上是一致的。本文针对岩土材料应变局部化的有限元新技术所存在的若干问题进行了详细的讨论,并给出了有关算例。     

组合导航系统多重衰减因子自适应估计算法比较研究

提出了多重衰减因子自适应估计卡尔曼滤波方法,用该方法对系统每个误差状态估计进行控制,提高滤波器的估计性能。仿真结果表明,新算法在系统噪声特性不准确的情况下,能够抑制卡尔曼滤波估计的发散,GPS/SINS组合导航精度比强跟踪滤波估计的精度高。这种算法推导形式简单,计算量小,适合在线运算。     

小波变换在湍流数值研究中的应用

小波变换具有时空双局部性特点,恰好适应了湍流特性。本文主要阐述了小波在湍流数值计算中的两大研究进展:一个是利用连续小波(高斯小波)可使L ap lace算子降阶的特性来求解N-S方程;另一个是正交小波与有限元法相结合的方法——相干涡模拟。指出它们的优缺点及其存在的问题,并对小波在湍流计算中的应用前景作了展望。     

耦合时空混沌的模糊混沌神经网络鲁棒自适应控制

将模糊逻辑系统和混沌神经网络结合起来,利用模糊逻辑系统的逼近能力和混沌神经网络的时空混沌行为,对模型未知的耦合时空混沌系统提出了一种模糊混沌神经网络自适应控制方案;同时考虑系统扰动、未建模动态特性和建模误差的影响,设计自适应补偿器,增强时空混沌系统控制的鲁棒性;并用Laypunov方法证明了该方案的稳定性;仿真验证了方案的有效性和鲁棒性。     

On Fractional Adaptive Control

Introducing fractional operators in the adaptive control loop, and especially in Model Reference Adaptive Control (MRAC), has proven to be a good mean for improving the plant dynamics with respect to response time and disturbance rejection. The idea of introducing fractional operators in adaptation algorithms is very recent and needs to be more established, that is why many research teams are working on the subject. Previously, some authors have introduced a fractional model reference in the adaptation scheme, and then fractional integration has been used to deal directly with the control rule. Our original contribution in this paper is the use of a fractional derivative feedback of the plant output, showing that this scheme is equivalent to the fractional integration, one with a certain benefit action on the system dynamical behaviour and a good robustness effect. Numerical simulations are presented to show the effectiveness of the proposed fractional adaptive schemes.     

Adaptive synchronization of chaotic systems with less measurement and actuation

We investigate the synchronization problem between identical chaotic systems only when necessary measurement(output) and actuation(input) are needed to be implemented by the adaptive controllers. A sufficient condition is derived based on the Lyapunov stability theory and Schur complementary lemma. Moreover, the theoretic result is applied to the Rikitake system and the hyperchaotic Liu system to show its effectiveness and correctness. Numerical simulations are presented to verify the results.     

基于贝叶斯参数优化的无模型自适应硅单晶直径控制

直拉硅单晶的生长过程涉及多场多相耦合与复杂的物理化学变化,其中工艺参数的波动是导致晶体直径不均匀的重要原因,如何实现工艺参数的控制以获得理想的、均匀的晶体直径具有重要的研究意义。本文分析现有控制方法存在不稳定以及控制效果不佳的问题后,提出基于贝叶斯参数优化的无模型自适应控制模型来控制硅单晶生长过程中的晶体直径。首先以坩埚上升速度与加热器的功率作为控制输入参数,晶体直径作为输出,搭建无模型自适应控制模型,并分析算法的稳定性。其次将控制模型进行仿真实验,发现硅单晶直径控制模型中不同的超参数设定会影响控制过程的迭代次数以及控制效果。最后,利用贝叶斯优化超参数的取值范围,并进行最终的仿真实验,结果表明,经贝叶斯参数优化后的控制模型计算快、迭代次数少,输出的晶体直径稳定,同时将生长工艺参数控制在实际生产要求范围内。因此,基于贝叶斯参数优化的无模型自适应控制实现了硅单晶直径均匀稳定的有效控制,具有结合工程背景的实际应用前景。     

A residual‐based Allen–Cahn phase field model for the mixture of incompressible fluid flows

The hydrodynamics of fluid mixtures is receiving more and more attention in many science and engineering applications. Within the techniques for dealing with front displacements and moving boundaries between different density and/or viscosity fluids, phase fields are a class of models in which a diffusive transition region is taken into account instead of a steep interface. Although these models have a physical motivation, they require the definition of extra parameters. In order to make it less parameter dependent, the classic Allen–Cahn phase field model is modified, exploring its similarities with residual‐based discontinuity‐capturing schemes, making the phase field equation dependent on its own residual. We solve the coupling between incompressible viscous fluid flow and the phase field advective–diffusive–reactive transport to simulate the main processes in interface tension and/or buoyancy driven problems. For the solution of the Navier–Stokes and transport equations, we use a stabilized finite element formulation. The implementation has been performed using the libMesh finite element library, written in C++ , which provides support for adaptive mesh refinement and coarsening. A chemical convection benchmark problem is used to validate the proposed model, and then we solve two bubble interaction problems. Copyright © 2014 John Wiley & Sons, Ltd.