摘要
针对门式起重机刚性支腿结构动态特性的复杂性和非线性,利用参数化有限元模型和BP神经网络,建立刚性支腿设计变量和最大动应力、弯曲动刚度及顶部最大动位移之间的映射关系.对于建立的神经网络模型,采用混合遗传算法(HGA)构造基于模糊动态罚函数的适应度函数引导遗传算法的搜索方向,寻求刚性支腿隔板、侧板的布置及尺寸最优化,并满足低应力、高固有频率及轻量化的要求.开发了某型号门式起重机刚性支腿多目标动态优化设计系统.应用结果表明,采用该优化方法能够有效地实现起重机刚性支腿的动态结构优化,显著提高了设计质量和效率.
Focused on the complexity and highly nonlinearity of the structural dynamics characteristic in the rigid landing leg of gantry crane, the parametric finite element model and the back propagation （BP） neural network were used to establish the mapping relationship between the design variables of rigid landing leg and the maximum dynamic stress, the bending dynamic stiffness, the maximum dynamic displacement on the top of rigid landing leg. The hybrid genetic algorithm （HGA） was adopted based on the established neural network model in order to find the layout optimization of the clapboards, the lateral plates and their sizes in rigid landing leg. The fitness function was constructed based on fuzzy dynamic penalty function to guide the searching direction. Then the requirements of low-stress, high natural frequency and lightweight were meeted. Multi-objective dynamic optimization design systems were developed for the rigid landing leg of a certain crane. Application results indicate that the dynamic structural optimization of rigid landing leg can be effectively conducted, and the design quality and efficiency are evidently improved.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2013年第1期122-130,共9页
Journal of Zhejiang University：Engineering Science
基金
上海市"创新行动计划"产学研联盟专项资助项目(2009C11060)
关键词
刚性支腿
参数化有限元
BP神经网络
混合遗传算法
动态罚函数
多目标优化设计
rigid landing leg
parametric finite element
BP neural network
hybrid genetic algorithm
dynamic penalty function
multi-objective optimization design
