基于动态RBF代理模型和进化算法的起重机主梁优化  

Optimization of crane main beam based on dynamic RBF agent model and evolutionary algorithm

作　　者：段雄 范小宁[1] DUAN Xiong;FAN Xiaoning(School of Mechanical Engineering,Taiyuan University of Science and Technology,Taiyuan 030024)

机构地区：[1]太原科技大学机械工程学院,山西太原030024

出　　处：《机械设计》2025年第3期86-94,共9页Journal of Machine Design

基　　金：山西省基础研究计划项目(20210302123212)。

摘　　要：针对基于有限元仿真模型的起重机结构优化计算成本在工程上难以接受的问题,文中结合差分进化算法和径向基代理模型提出一种基于动态径向基代理模型的全局优化策略。该策略在优化过程中通过局部开发最优解和全局探索误差最大区域的加点策略构造动态径向基代理模型,并以约束函数模型的预测误差和目标函数下降程度构建优化终止条件,保证优化的全局收敛性和最优解处的模型精确性。通过数值算例和工字梁优化算例进行验证,该方法不仅能够获得全局最优解,而且明显减少了对原函数的调用次数,显著提高了优化效率。最后,结合桥式起重机桥架的有限元分析,将此方法用于解决起重机主梁优化问题。结果显示:在满足约束的条件下,主梁横截面面积减小了约22.36%,并且降低了大量的计算成本,提高了优化效率,解决了智能群算法与起重机结构有限元模型直接结合进行优化的昂贵计算成本问题。Since the computational cost on crane structure optimization based on the finite-element simulation model is unaffordable in engineering,in this article a global optimization strategy based on the dynamic RBF agent model is proposed by combining the differential evolutionary algorithm and the RBF agent model.The dynamic RBF agent model is constructed during the optimization process through the adding point strategy of locally exploiting the optimal solution and globally exploring the region with the largest error.Meanwhile,the optimized termination condition is identified through the prediction error of the constraint function model and the degree of decline of the objective function,which ensures the optimization’s global convergence and the model’s accuracy at the optimal solution.The numerical and I-beam optimization examples verify that this method not only achieves the global optimum solution,but also significantly reduces the number of calls to the original function and greatly improves the optimization efficiency.Finally,combined with the finite-element analysis on a bridge crane bridge,this method is used for main beam optimization.The results show that while the constraints are met,the main beam’s cross-sectional area reduces by about 22.36%,thus saving a large amount of computational cost and improving the optimization efficiency.As a result,the problem of the huge computational cost due to a direct combination of the intelligent swarm algorithm and the finite-element model of the crane structure for optimization is addressed.

关 键 词：起重机主梁 动态径向基代理模型 差分进化算法 加点策略 

分 类 号：TH215[机械工程—机械制造及自动化]