BP-NSGA和FEM相结合优化万能型钢轧机机架圆角  

Housing fillet optimization of universal section mill by BP-NSGA combined with FEM

作　　者：马劲红[1] 陈伟[1] 张文志[2] 李娟[2] 

机构地区：[1]河北联合大学冶金与能源学院,河北唐山063009 [2]燕山大学机械工程学院,河北秦皇岛066004

出　　处：《锻压技术》2014年第6期137-141,共5页Forging & Stamping Technology

基　　金：唐山市科技发展计划资助项目(08360201A-1)

摘　　要：万能型钢轧机机架圆角直接影响机架的应力分布、纵向和横向变形,从而影响机架的强度、刚度、寿命、H型钢的精度和尺寸。根据万能型钢轧机机架的受力特点和几何对称性,建立机架的有限元模型。利用有限元软件ANSYS分析机架圆角对其应力分布及位移的影响。把改进的带精进策略的非支配分类遗传算法(NSGA)和有限元相结合,建立新的优化方法。并用BP神经网络建立非支配分类遗传算法(NSGA)的适应度函数。以机架的纵向位移和体积为目标函数,利用这种新的优化方法和ANSYS自带的零阶和一阶优化方法优化万能型钢轧机机架圆角,与ANSYS自带的零阶和一阶优化方法相比,这种新的优化方法取得了良好的优化效果。The housing fillet of universal section mill directly influents the stress distribution and the longitudinal and lateral deformation of housing, thus affects its intensity, stiffness, service life and H-beam precision and size. According to the stress features and geometric symmetry of the housing fillet of the housing fillet of universal section mill, the FEM model of housing was set up. The influence of housing fillet on the stress distribution and displacement was analysed by FEM software ANSYS. By the improved Non-dominated Sorting Genetic Algorithms （NSGA） with elite strategy combined with FEM, a new kind of optimization method was established. BP neural network was used to establish the fitness function of NSGA. Taking longitudinal displacement and volume of housing as objective function, this new kind of optimization method and the zero-order and first -order optimization method with ANSYS were used to optimize the housing fillet of universal section mill. Compared with the zero-order and first-order optimization method with ANSYS, the optimization result of this new optimization method is better.

关 键 词：万能型钢轧机 机架圆角 改进的NSGA BP神经网络 FEM 

分 类 号：TG333.13[金属学及工艺—金属压力加工]