基于BP神经网络的管材数控弯曲多参数优化方法研究

发布时间：2018-03-04 16:34

本文选题：管材数控弯曲　切入点：BP神经网络　出处：《锻压技术》2015年06期 　论文类型：期刊论文

【摘要】：针对管材数控弯曲成形过程中多工艺参数耦合的特点,基于BP神经网络,结合多目标优化算法研究了弯曲成形工艺参数的优化方法。采用ABAQUS对管材数控弯曲过程进行有限元仿真,并实验验证了结果的准确性。基于MATLAB平台,以芯棒直径、芯棒伸出量、防皱块与管材间摩擦系数等主要工艺参数为优化对象,以外壁减薄率、内壁增厚率(起皱)为优化目标,通过验证的数值模型获得样本数据,利用BP神经网络建立了优化对象和优化目标之间的映射关系,并采用多目标优化算法进行寻优求解,最后通过数值仿真实验验证了优化方法的准确性。结果表明:薄壁管数控弯曲有限元数值模拟结果与实验数据吻合较好,可以为神经网络提供准确可靠的训练样本;BP神经网络结合多目标优化算法可以有效地对弯曲工艺参数进行优化;优化的工艺参数有效地改善了弯曲管材内侧起皱和外侧减薄。
[Abstract]:In view of the characteristics of coupling of multi-process parameters in tube NC bending process, BP neural network is used. Combined with multi-objective optimization algorithm, the optimization method of bending process parameters is studied. The finite element simulation of tube numerical control bending process is carried out by using ABAQUS, and the accuracy of the results is verified by experiments. Based on MATLAB platform, the diameter of mandrel and the amount of mandrel outstretched are used. The main technological parameters such as friction coefficient between anti-wrinkle block and pipe are optimized, the thinning rate of outer wall and the wrinkling rate of inner wall are the optimization objectives. The sample data are obtained by the verified numerical model. The mapping relationship between the optimization object and the optimization object is established by using BP neural network, and the multi-objective optimization algorithm is used to solve the optimization problem. Finally, the accuracy of the optimization method is verified by numerical simulation experiments. The results show that the numerical simulation results of numerical control bending of thin-walled tubes agree well with the experimental data. The BP neural network can provide accurate and reliable training samples for the neural network and the multi-objective optimization algorithm can effectively optimize the bending process parameters, and the optimized process parameters can effectively improve the inner wrinkling and outer thinning of the curved pipe.
【作者单位】：武汉理工大学现代汽车零部件技术湖北省重点实验室;汽车零部件技术湖北省协同创新中心;
【基金】：中央高校基本科研业务费专项资金资助项目(2014-IV-042)
【分类号】：TG306

【参考文献】