基于黎曼几何的非线性振子动力学分析递推解析算法

基于黎曼几何和变分原理，推导了黎曼流形上非线性耗散动力系统的二阶微分动力学方程，并运用流形收缩的概念将动力学方程离散化，进而建立了相应的递推求解格式。选取3个自治非线性阻尼振子系统，分别采用递推解析算法和龙格库塔法求解微分动力学方程，并比较分析了不同的时间步长下两种算法的计算耗时。结果表明，与龙格库塔法相比，基于黎曼几何的递推算法不仅能得到每一时步的解析表达式，而且计算耗时短，计算效率高。基于黎曼流形的动力学方程递推算法为非线性动力学系统的解析求解提供了新思路。     

基于遗传算法的设计地震反应谱标定方法

基于自然进化原理的遗传算法是一种解决复杂非线性规划问题十分有前景的方法。对遗传算法中一些有代表性的算法作了简要分析和评价。介绍了用地面峰值加速度(PGA)、峰值速度(PGV)和峰值位移(PGD)三个参数标定设计地震反应谱的模型。针对此分段非线性函数规划问题,运用MATLAB编写遗传算法程序进行了计算。计算表明遗传算法在全局优化、非线性优化、多参数优化等方面表现出了传统算法无法比拟的优势。     

多种群遗传算法在PBX本构模型参数识别中的应用

利用多种群并行结构对标准遗传算法SGA进行并行化处理,引入移民算子和精华种群形成多种群遗传算法MPGA,并设计了自适应交叉和变异概率对算法的收敛速度进行改进。结合ABAQUS软件和改进的多种群遗传算法,建立了材料本构模型参数识别方法。采用该方法对PBX炸药黏弹性损伤本构模型参数进行了模拟识别,并同基于标准遗传算法的参数识别方法进行了比较。结果证明,基于改进多种群遗传算法IMPGA的方法对克服算法未成熟收敛有显著的效果,识别结果更稳定。同时该方法的收敛速度更快,寻优能力更强,适合复杂非线性问题的优化,此方法可以被应用到其他材料本构模型的参数识别中。     

部分输入未知条件下结构动力复合反演的分解算法

在充分利用部分输入已确知而部分输入未知的激励特性的基础上，提出了结构动力复合反演的分解算法，该算法从源头上消除了迭代过程中参数识别与荷载反演的相互影响，降低了问题的计算规模。对于线性参数系统，该算法不经过任何迭代计算即可一次性完成结构参数识别及荷载反演。将其与松弛法结合，可解决非线性参数系统的识别问题，与文献[4]的方法比较，其收敛速度有显著提高。     

基于遗传算法的水轮机振动参数识别

通过对水轮机现场测试数据的频域分析，建立了测点位移响应的参数正模型。并基于遗传算法，建立了在时域内识别结构振动参数的数值方法，采用该方法对水轮机位移响应正模型的待定参数识别进行了研究。计算结果表明，所建立的基于遗传算法的参数识别方法具有良好的稳定性、抗观测噪声能力和较高的识别精度，能够反演出比较符合实际情况的位移响应振动模型。这就为荷载识别、结构优化和预报减震效果等后续工作提供了可靠的实施前提。     

基于一种免疫算法的结构动态参数识别

受生物免疫系统的启发提出了一些基于免疫概念的优化算法。该方法有效地避免了传统遗传算法易于出现早熟、搜索效率低及不能很好保持个体多样性等问题。本文将免疫算法应用于结构动态参数识别领域,通过其在VC环境下调用ANSYS有限元程序中的APDL语言,并在后期引入加速循环操作,从而实现了对结构动态参数快速而精确的识别,仿真算例表明其正确性、有效性和较好的抗观测噪声能力。该方法特别适合于对大型复杂结构的参数识别计算。     

一种非线性系统周期解的延拓算法

提出了一种非线性系统周期解的延拓算法。指出了非线性系统周期解在分岔点处由于雅可比矩阵奇异而导致一般延拓方法延拓失败问题;然后基于推广的打靶法的思想,将普通延拓算法推广,提出了一种用于周期解延拓的算法。对于非线性动力系统,该算法可以在已知某一参数下的周期解的基础上,求解出在一定参数范围内非线性动力系统的解随参数的连续变化情况。应用该方法对非线性柔性转子-轴承系统的周期解与参数的依赖关系进行了求解,验证了方法的有效性。     

二阶非定常多宗量热传导反问题的正则解

引入Bregman距离函数及其加权函数作为正则项，应用Tikhonov正则 化方法，对二阶非定常多宗量热传导反问题进行求解. 利用测量信息和计算信息构造最小二 乘函数，将多宗量反演识别问题转化为一个优化问题. 空间上采用8节点等参元进行离散， 时域上采用时域精细算法进行离散，建立了二阶非定常多宗量热传导问题的有限元正/反演数 值模型. 该模型不仅考虑了非均质和参数分布的影响，而且也便于正反演问题的敏度分析， 可对导热系数和边界条件等宗量进行有效的单一和组合识别. 给出了相关的数值验证，对信 息测量误差以及不同正则项的计算效率作了探讨. 数值结果表明，该方法能够对二阶非定常 多宗量热传导反问题进行有效的求解，并具有较高的计算精度.     

基于改进萤火虫算法的移动车辆参数识

提出了一种基于萤火虫优化算法的移动车辆参数直接识别方法.采用四自由度双轴十二参数车辆模型和欧拉梁有限元模型建立了车桥耦合系统的动力方程,并利用Newmark直接积分法求解了系统的动力响应.通过引入一种局部搜索策略和末位淘汰机制改进了萤火虫优化算法的收敛速率,并提高了识别结果的精度.本文方法仅利用车辆振动的竖向加速度响应测量就能进行移动车辆参数的识别.数值算例表明,改进的萤火虫优化算法可以准确地识别出车辆的质量、悬挂刚度和阻尼等参数,并且对测量噪声不敏感.     

求解非线性动力系统周期解大范围收敛方法

对于多自由度非线性动力系统,提出一种求解周期解的大范围收敛方法,这种算法对处理非线性动力系统有较强的功能。结合数值延拓算法,为求解具有系统参数的非线性动力系统在整个系统参数范围内的周期解提供了有效的方法。     

基于遗传算法的爆炸冲击荷载参数识别方法

基于改进的遗传算法 ,建立了根据测试系统动力响应观测数据反演爆炸冲击荷载参数的数值方法。遗传算法为解决反问题的不适定性提供了强有力的手段。数值模拟结果表明 ,所提出的爆炸冲击荷载参数随机反演方法具有全局搜索能力 ,并且具有良好的抗观测噪音能力。当测试系统的观测相对误差为 10 %时 ,参数反演结果的误差小于 8% ,所建立的参数反演方法具有良好的鲁棒性。     

Effects of the method of identification of the diffusion coefficient on accuracy of modeling bound water transfer in wood

An alternative approach to determining the bound water diffusion coefficient is proposed. It comprises a method for solving the inverse diffusion problem, an improved algorithm for the bound-constrained optimization as well as an alternative submodel for the diffusion coefficient’s dependency on the bound water content. Identification of the diffusion coefficient for Scots pine wood (Pinus sylvestris L.) using the proposed inverse approach is presented. The accuracy of predicting the diffusion process with the use of the coefficient values determined by traditional sorption methods as well as by the inverse modeling approach is quantified. The similarity approach is used and the local and global relative errors are calculated. The results show that the inverse method provides valuable data on the bound water diffusion coefficient as well as on the boundary condition. The results of the identification can significantly improve the accuracy of mass transfer modeling as studied for drying processes in wood.     

计算不确定结构系统静态响应的一种可靠方法

不确定性广泛存在于工程结构分析和设计过程之中，不能简单地予以忽略。目前，概率方法、模糊方法和区间方法是不确定性建模的三种主要方法。本文把具有不确定性的结构材料参数、几何参数和所受外力用区间数描述，通过求解线性区间方程组准确地计算了结构静态响应。计算结果易于扩张是区间计算的一个主要缺陷，本文提出了一种有效避免这一问题的方法。该方法把区间函数的计算和区间线性方程组的求解转化为相应的全局优化问题，来确定解中的每个区间元素的边界值，并采用一种智能性算法(实数编码遗传算法)来求解这些全局优化问题。本文首先采用数学和结构分析算例对该方法的正确性和有效性进行了验证，然后把该方法与有限元方法相结合计算不确定结构系统的响应范围，并和求解同类问题的方法进行了比较。     

A LEVEL-VALUE ESTIMATION METHOD FOR SOLVING GLOBAL OPTIMIZATION

A level-value estimation method was illustrated for solving the constrained global optimization problem. The equivalence between the root of a modified variance equation and the optimal value of the original optimization problem is shown. An alternate algorithm based on the Newton's method is presented and the convergence of its implementable approach is proved. Preliminary numerical results indicate that the method is effective.     

Inverse parameter identification of cohesive zone model for simulating mixed-mode crack propagation

Inverse analysis is widely applied to the identification of material properties or model parameters. In order to improve the computational efficiency of the inverse method based on the genetic algorithm, an interpolation scheme upon the response surface constructed by the finite element simulation has been adopted in this paper. Meanwhile, a gradual homogenization treatment scheme has also been presented to improve the convergence of the inverse method based on the Kalman filter algorithm. Both methods are proven effective in dealing with the single-objective inverse problem. However, literature studies show that the adoption of multiple types of experimental information is useful to improve the accuracy of inverse analysis. In this case, it turns into a multiple-objective inverse problem. Our practice proved that the above-mentioned two methods might not yield a proper result if the sensitivity issue of different types of information is not considered. Therefore, another multi-objective inverse method, in combination of the above two optimization algorithms and a weight-estimating scheme that can consider such sensitivity, has been further presented. Finally, by using a mixed-mode crack propagation simulation and two types of experimental information (loading-displacement response curve and crack path profile), the parameters of the cohesive zone model were inversely identified and its simulation results are in good agreement with the experiment.     

基于CMDLAC指标和遗传算法的结构损伤定位研究

提出了基于综合多损伤定位保证准则(CMDLAC)指标和遗传算法(GA)的结构损伤定位方法。基于结构损伤标量模型,结合灵敏度分析提出了基于频率和振型变化的CMDLAC指标,将损伤识别问题转化为约束优化问题,最后以CMDLAC指标为优化目标函数应用遗传算法求解实现了结构的损伤定位。在考虑测试自由度不完备和测量噪声等因素影响的情况下,对桁架结构损伤的数值模拟分析表明了方法的有效性,验证了CMDLAC指标定位损伤和抵抗噪声的能力,同时指出其具有较强的实际可操作性。     

Parameter identification for improved viscoplastic model considering dynamic recrystallization

The successful application of viscoplastic model considering dynamic recrystallization depends on how well the parameters are identified. However, it is difficult to obtain satisfactory parameters using conventional parameter identification methods. The reasons are due to difficulties in obtaining homogeneous deformation, high complexity of physical process described by the model and large number of parameters. In this paper, the material parameters are identified by inverse analysis. Global information on objective function is firstly studied by an improved uniform random sampling method; secondly, a hybrid global optimization method, which combines the genetic algorithm, the Levenberg–Marquardt algorithm, the augmented Gauss–Newton algorithm and the flexible tolerance method, is constructed and an inverse analysis numerical procedure, which combines the proposed optimization method with the finite element analysis, is proposed; at last, a set of satisfactory material parameters for 26Cr2Ni4MoV is obtained by the proposed inverse analysis numerical procedure.     

Identification of vibration loads on hydro generator by using hybrid genetic algorithm

Vibration dynamic characteristics have been a major issue in the modeling and mechanical analysis of large hydro generators. An algorithm is developed for identifying vibration dynamic characteristics by means of hybrid genetic algorithm. From the measured dynamic responses of a hydro generator, an appropriate estimation algorithm is needed to identify the loading parameters, including the main frequencies and amplitudes of vibrating forces. In order to identify parameters in an efficient and robust manner, an optimization method is proposed that combines genetic algorithm with simulated annealing and elitist strategy. The hybrid genetic algorithm is then used to tackle an ill-posed problem of parameter identification, in which the effectiveness of the proposed optimization method is confirmed by its comparison with actual observation data.The project supported by the National Natural Science Foundation of China (10472025). The English text was polished by Yunming Chen.     

Erratum to: Parameter identification of electronic throttle using a hybrid optimization algorithm

Aiming at the problems in parameter identification of an electronic throttle, this paper proposes a novel hybrid optimization algorithm to search the optimal parameter values of the plant. The parameter identification of an electronic throttle is considered as an optimization process with an objective function minimizing the errors between the measurement and identification, and the optimal parameter values of the plant are searched by using a hybrid optimization algorithm. The proposed hybrid optimization algorithm, effective combination of parallel chaos optimization algorithm (PCOA) and simplex search method, preserves both the global optimization capability of PCOA and the accurate search ability of simplex search method. Simulation and experiment results have shown the good performance of the proposed approach.