摘要
为得到钣金成形最佳工艺方案,结合有限元方法将设计转化为特定目标和约束的待优化问题;针对成形模拟易受干扰和强非线性特点,提出采用逐次逼近模型,分解复杂设计函数为显式简单函数组合,进行递进全局寻优,避免出现局部最优解和过程发散现象;并利用动态多项式序列响应面方法构造目标和约束的简单近似响应面,用以去除噪声干扰及最大程度减少精确分析计算量。该方法无需求解复杂函数敏度,经实例验证具有很高的精度和效能。
In order to get optimum solution for sheet metal forming design, an optimization methodology was developed, which was based on FE simulation and satisfied some given design requirements and constraints. Sequential approximation models that can decrease highly non-linear effects were carried out to decompose complex design functions into combinations of explicit functions and to find a global optimum. Dynamic Sequential Response Surface Methodology (DSRSM) based polynomial was adopted to build approximate response surfaces of object and constraints so as to eliminate computational noise and to help to converge faster. The results of examples demonstrate that this method is efficient, accurate and robust for sheet metal forming process design.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2005年第4期307-310,共4页
China Mechanical Engineering
基金
浙江省重大科技攻关项目(021101557)
关键词
钣金成形
优化
逐次逼近
响应面方法
有限元模拟
sheet metal forming
optimization
sequential approximation
response surface method
FE simulation