基于多目标元胞遗传算法的桁架结构优化设计

Optimal Design of Truss Structure Based on Cellular Genetic Algorithm for Multi-objective Optimization

为了使桁架结构的设计更符合工程实际情况,建立了同时最小化结构总质量和节点位移的多目标优化模型。采用离散实数编码方式,使多目标元胞遗传算法适用于优化模型的求解,同时改进了算法的交叉变异算子,使算法保持较好的全局探索和局部寻优能力。运用该算法对经典空间桁架优化问题进行求解,并与NSGA-II的优化结果进行对比分析,结果表明该算法获得的Pareto前端更加均匀,解的精度更高,极端点值域更广,是解决此类离散优化问题的有效算法。

In order to make the design of truss structure more accord with the actual engineering,an optimization model was established with the goals of the minimum structural mass and the nodal displacement.Based on discrete real number coding,improved simulated binary crossover and polynomial mutation,cellular genetic algorithm for multi-objective optimization（ M OCell） was successfully applied to the optimal model,integrating excellent ability of global exploration and local exploitation. To evaluate the algorithm,a comparison with NSGA-II on the optimization problem of the well-known space truss was conducted. The results indicated that the pareto front of M OCell had more uniform distribution,higher accuracy and wider range of the extreme points. It could solve this kind of discrete optimization problems effectively.