工程结构混合离散变量优化的模拟退火方法

本文针对工程结构优化设计中普遍存在的混合离散变量的现象，探索了应用模拟退火算法直接获得最优解的方法，从而不再对离散变量作“规格化”后处理。文中根据混合离散变量的特点，提出了几种邻域状态的产生函数和迭代方案，给出了适宜的模拟退火过程的冷却进度表。算例表明该方法能有效地解决工程结构混合离散变量的优化问题。     

基于SA算法的行星齿轮减速器离散变量优化设计

考虑制造工艺要求,将所有设计变量均视为离散变量,包括一般离散变量和伪离散变量,并就这两种情况下状态产生函数的设计原理进行深入研究,解决了将模拟退火算法用于离散变量函数优化的关键技术问题,介绍了一种基于模拟退火算法的离散变量函数优化的新方法。行星齿轮传动中各齿轮的齿数受传动比条件、同轴条件和装配条件的限制而不能任意取值,齿轮的模数也要受国家标准的制约只能取一些离散值,用以数学规划理论为基础的经典约束优化方法求解效果很差,用基于模拟退火算法的离散变量优化设计方法则可以方便快捷地获得满足各方面要求的最优设计方案。     

拓扑优化方法在复合材料设计中的研究进展

按照设计变量,叙述了连续变量和离散变量拓扑优化设计的一些常用算法,其中包括均匀化方法、变密度法、变厚度法、移动渐进算法、模拟退火法、遗传算法、相对差商法和Tabu搜索法,并对各种方法的优缺点进行了比较;对二维和三维复合材料的拓扑学优化设计研究现状和方法进行了阐述;提出了拓扑优化设计复合材料的未来研究方向.     

离散变量结构优化的拟满应力设计方法

本文以满应力设计思想为基础,提出了适用于离散变量结构优化设计计算的拟满应力设计方法。该方法能直接计算具有应力约束和截面尺寸约柬的离散变量结构优化设计问题,也能处理同时具有稳定性约束和位移约束的多工况、多约束、多变量的离散变量结构优化设计问题。算例结果表明,拟满应力设计方法对于离散变量结构优化计算是非常有效的。     

离散变量桁架结构拓扑优化设计的混合算法

将相对差商法和混沌优化结合起来,形成求解离散变量桁架结构拓扑优化设计的混合算法。利用相对差商法可以对离散变量快速寻优的特点,及混沌变量的全局遍历性,可以有效地跳出局部最优解,达到拓扑优化全局寻优的目的。通过采用和准最优解的对比及几何稳定性的判断等辅助性技术,降低了重分析次数。同时,高效的重分析方法的结合,提高了求解的效率,也避免了拓扑优化问题中求解的一些困难。算例表明,该算法对于离散变量的拓扑优化设计问题是快速有效的。     

混合离散变量多目标结构模糊优化及FD解法

本文建立了混合离散变量多目标结构模糊优化设计的数学模型：提出了模糊寻优区间、模糊优越混合离散解集和模糊可行集的概念；     

离散复合形法的改进及应用研究

在工程设计中,复合形法是求解约束非线性规划的一种直接方法.它计算简单、适用性强,是解决工程技术问题的常用方法.通常情况下,复合形法多用于连续变量优化设计问题的求解,但在工程实际中还存在大量的离散变量优化问题.为此,我们在连续变量复合形算法的基础上,采用平均值舍入法获取离散点,并改进了复合形迭代过程和停机准则,使之成为一种能直接求解离散变量优化解的方法.几个典型工程离散变量优化设计算例的计算结果表明,我们提出的方法十分有效.     

离散变量结构优化设计的拟满应力遗传算法

以力学准则法为基础,提出了一种求解离散变量结构优化设计的拟满应力方法;这种方法能直接求解具有应力约束和几何约束的离散变量结构优化设计问题。通过在遗传算法中定义拟满应力算子,建立了一种离散变量结构优化设计的混合遗传算法拟满应力遗传算法。算例表明:这种混合遗传算法对于离散变量结构优化设计问题具有较高的计算效率。     

基于遗传算法的混合离散变量结构优化设计

本文研究了适于解决混合离散变量结构优化设计的遗传算法，给出了混合离散变量的遗传基因表达模式和适值函数建立的简便方法，计算实例表明该算法是有效的。     

智能优化平台的并行实现及在光学设计中的应用研究

构建了一个智能优化平台,包括粒子群算法、蚁群算法、遗传算法、模拟退火算法、混沌算法和复合形法。该平台包括4个模块,分别是优化问题选择,优化参数设置,优化过程显示以及优化结果输出。将该优化平台应用在RGB LED混合白光的优化设计中,对混合照明模块中发光效率和显色指数进行优化,给出了混合照明模块中发光效率和显色指数的计算方法。     

Engineering optimization using a real-parameter genetic-algorithm-based hybrid method

A hybrid optimization algorithm which combines the respective merits of the genetic algorithm and the simulated annealing algorithm is proposed. The proposed algorithm incorporates adaptive mechanisms designed to adjust the probabilities of the cross-over and mutation operators such that its hill-climbing ability towards the optimum solution is improved. The algorithm is used to optimize the weight of four planar or space truss structures and the results are compared with those obtained using other well-known optimization schemes. The evaluation trials investigate the performance of the algorithm in optimizing over discrete sizing variables only and over both discrete sizing variables and continuous configuration variables. The results show that the proposed algorithm consistently outperforms the other optimization methods in terms of its weight-saving capabilities. It is also shown that the global searching ability and convergence speed of the proposed algorithm are significantly improved by the inclusion of adaptive mechanisms to adjust the values of the genetic operators. Hence the hybrid algorithm provides an efficient and robust technique for solving engineering design optimization problems.     

Single and multiobjective structural optimization in discrete-continuous variables using simulated annealing

A multivariable optimization technique based on the Monte-Carlo method used in statistical mechanics studies of condensed systems is adapted for solving single and multiobjective structural optimization problems. This procedure, known as simulated annealing, draws an analogy between energy minimization in physical systems and objective function minimization in structural systems. The search for a minimum is simulated by a relaxation of the statistical mechanical system where a probabilistic acceptance criterion is used to accept or reject candidate designs. To model the multiple objective functions in the problem formulation, a cooperative game theoretic approach is used. Numerical results obtained using three different annealing strategies for the single and multiobjective design of structures with discrete-continuous variables are presented. The influence of cooling schedule parameters on the optimum solutions obtained is discussed. Simulation results indicate that, in several instances, the optimum solutions obtained using simulated annealing outperform the optimum solutions obtained using some gradient-based and discrete optimization techniques. The results also indicate that simulated annealing has substantial potential for additional applications in optimization, especially for problems with mixed discrete-continuous variables.     

Monte Carlo structural optimization in discrete variables with annealing algorithm

The paper describes the basic ideas of Monte Carlo annealing algorithms for structural optimization with discrete design parameters. The algorithm generates randomly a set of design parameters, with probability depending on the objective function and given by the Boltzmann–Gibbs distribution. In this model the search for the global minimum is simulated by a relaxation process of the statistical mechanical system with the Hamiltonian proportional to the objective function. The rate of the convergence of the method and its dependence upon the annealing probability are discussed. Numerical implementation of the method for the weight optimization of the ten-bar planar cantilever truss is presented. The results of numerical simulation are compared with those obtained by the dual methods. The principal conjecture is that the method is fairly efficient and has great potential for applicaton in engineering design.     

基于梯度优化方法的钢结构标准截面选型优化设计

实际的工程应用中,钢框架的基本构件大多是根据钢结构设计规范要求,从标准型钢库中选取,所组成的框架结构的截面尺寸非连续变化。因此,钢结构截面优化设计是典型的离散设计变量优化问题。若采用基于启发式的算法(如遗传算法等)进行求解,当可选截面类型较多时,其计算量巨大,求解效率低下。该文通过引入高维拉格朗日插值函数对该离散设计问题进行连续化,建立了可采用梯度优化方法进行求解的钢结构标准截面选型设计模型,并且使得连续化以后的设计变量个数大幅度减少。对给定截面类型种数为2n个的可选截面集合,其设计变量只需n个即可。具体算例表明:与基于遗传算法的优化方法相比,该方法的计算效率提高1~2个数量级,并且在结构性能基本相当的情况下,得到的型钢种类更少,便于工程应用。     

Solving systems of non‐linear equations with both free and positive variables

A numerical method is presented for solving systems of non‐linear equations that contain some variables that are strictly positive and others that have no restriction on sign. Naturally positive variables arise frequently when modelling the behaviour of engineering systems, such as physical dimension, concentration of a chemical species, duration of an event, etc. When modelling systems of this type, it is also common to introduce additional variables that are not restricted in sign, such as stresses, displacements, velocities, accelerations, etc. Many numerical methods may experience performance difficulties due to the existence of spurious solutions which have negative components for one or more of the positive variables. Recently, the monomial method has been developed as an effective tool for systems with variables that are all strictly positive. This paper presents a hybrid method, combining the monomial method and Newton's method, for systems containing both types of variables. It is demonstrated that this hybrid method can be more effective in solving systems of equations with both positive and free variables than either method alone. Basins of attraction constructions are presented as a demonstration of the effectiveness of the hybrid method as applied to the design of a civil engineering frame structure. Copyright © 1999 John Wiley & Sons, Ltd.