论文部分内容阅读
基于模拟退火算法提出了一种结构优化设计的校正多级多点模拟退火算法(CMLPSA)。该算法从随机生成的一组试算点选择每个候选设计,实现一种高效的搜索策略。因而,CMLPSA是处理一组设计而不是一个单一的试算点。对可行和不可行的中间设计采用多点策略,即对可行情况,优化变量的变动是随着成本函数的当前变化率而变化;不可行情况下,在当前退火周期中生成的每个可行的试算点附近,进行第四阶近似直线搜索。此外,CMLPSA包括多级退火策略,通过同时(整体级)或逐个(局部级)变动所有的设计变量生成试算点。基于优化过程中的趋势,执行全局或局部搜索。数值算例验证了CMLPSA算法的有效性,可应用于结构优化。
Based on the simulated annealing algorithm, a modified multi-level simulated annealing algorithm (CMLPSA) with structural optimization design is proposed. The algorithm selects each candidate design from a randomly generated set of trial points to achieve an efficient search strategy. Thus, CMLPSA handles a set of designs instead of a single trial point. A multi-point strategy is used for both feasible and infeasible intermediate designs. That is, where feasible, the variation of the optimization variables varies with the current rate of change of the cost function. If feasible, each feasible one generated in the current annealing cycle Near the trial point, the fourth-order approximate straight line search. In addition, CMLPSA includes a multi-stage annealing strategy to generate trial points by varying all design variables at the same time (overall level) or one by one (local level). Based on trends in the optimization process, perform a global or local search. Numerical examples verify the effectiveness of CMLPSA algorithm and can be applied to structural optimization.