锻压技术

基于RBF神经网络的强力旋压连杆衬套成形质量预测研究

英文标题：Study on forming quality prediction of connecting rod bushing by power spinning forming based on RBF neural network

作者：高立樊文欣马学军李瑰

单位：中北大学

关键词：强力旋压工艺参数成形质量 RBF神经网络

分类号：TH161；TG376

出版年,卷(期):页码：2015,40(9):134-138

摘要：

为了实现对强力旋压连杆衬套成形质量（壁厚差和扩径量）的预测，进而对工艺参数进行优化，利用MATLAB人工神经网络工具箱，建立了强力旋压工艺参数与成形质量的RBF神经网络模型。基于减聚类算法改进的K-means学习算法，用模拟实验所得数据对神经网络进行训练，进而对旋压成形质量进行预测，通过与实测值对比，发现所建神经网络模型预测性能良好，实现了RBF神经网络在强力旋压领域的成功应用，与原始K-means学习算法训练的RBF神经网络和BP神经网络所建模型比较，发现改进K-means学习算法训练的RBF神经网络预测模型拥有更优的性能。该模型不仅可以为工艺参数的优化提供参考，还能缩短工艺参数的优化周期和减少实际实验的成本。

In order to predict the forming quality（wall-thickness-difference and inner diameter expending quantity） of connecting rod bushing by power spinning forming and optimize the process parameters, a RBF neural network model showing the relationship between process parameters and forming quality was established by neural network toolbox in MATLAB. Based on the improved K-means algorithm, the RBF neural network was trained by the experiment data and forming quality was predicted. Comparing with the measured data, it is found that the model shows high accuracy. Therefore, RBF neural network can be used in the power spinning field. Through comparing BP neural network and with RBF neural network trained by original K-means algorithm, it is indicated that RBF neural network has better accuracy and adaptability. This method can not only provide reference for optimization of process，but also shorten the period of process parameter optimization and save the cost on experiments.

基金项目：

高立（1991-），男，硕士研究生

参考文献：

［1］雷萍.小波神经网络技术在齿轮箱轴承故障诊断中的应用[D].兰州：兰州理工大学，2009.Lei P. Application of Wavelet Neural Network to Gear Box Bearings Fault Diagnosis[D].Lanzhou:Lanzhou University of Technology,2009.
［2］张剑，汤禹成.基于BP神经网络响应曲面的筒形件强力旋压工艺参数优化研究[J].锻压装备与制造技术，2007，41（1）：71-75.Zhang J,Tang Y C.Process parameters optimization of cylindrical workpieces based on BP neural network response surface methodology[J].China Metalforming Equipment & Manufacuring Technology, 2007,41(1):71-75.
［3］冯志刚，樊文欣，赵俊生，等.基于BP神经网络的强力旋压成形连杆衬套力学性能预测[J].热加工工艺，2014，43(5) ：89-91.Feng Z G,Fan W X,Zhao J S,et al.Prediction of mechanical property of power spinning forming connecting rod bushing based on BP neural network[J]. Hot Working Technology, 2014,43(5):89-91.
［4］罗亚军，何丹农，张永清，等.人工神经网络在塑性成形领域中的应用研究[J].锻压技术，2001，26(5)：46-49.Luo Y J,He D N,Zhang Y Q,et al.Application study on ANN in field of plastic forming[J].Forging & Stamping Technology, 2001,26(5):46-49.
［5］Haykin Simon. Neural Networks: A Comprehensive Foundation[M].New Jersey: Prentice-Hall, 1999.
［6］田银，谢延敏，孙新强，等.基于鱼群RBF神经网络和改进蚁群算法的拉深成形工艺参数优化[J].锻压技术，2014，39(12)：129-136.Tian Y,Xie Y M,Sun X Q,et al.Process parameters optimization of deep drawing based on fish RBF neural network and improved ant colony algorithm[J]. Forging & Stamping Technology, 2014,39(12):129-136.
［7］闻新，周露，李翔，等. MATLAB神经网络仿真与应用[M].北京:科学出版社，2003.Wen X,Zhou L,Li X,et al.MATLAB Neural Network Simulation and Application[M].Beijing:Science Press,2003.
［8］张俊明，刘军，俞小峰，等.一种RBF神经网络在某冷连轧机组轧制力计算中的组合应用[J].塑性工程学报，2008，15(1)：136-137.Zhang J M,Liu J,Yu X F,et al.Application of the combination of a sort of RBF neural network in roll force calculation of certain tandem cold mill[J].Journal of Plasticity Engineering, 2008,15(1):136-137.
［9］Chill S L.Fuzzy model identification based on cluster estimation[J].Journal of Intelligent and Fuzzy System,1994,2(3):1240-1245.
［10］庞振，徐蔚鸿.一种基于改进k-means的RBF神经网络学习方法[J].计算机工程与应用,2012,48(11)：161-163.Pang Z,Xu W H.Learning algorithm for RBF neural networks based on improved k-means algo-rithm[J].Computer Engineering and Applications,2012,48(11):161-163.
［11］张军峰，胡寿松.基于一种新型聚类算法的RBF神经网络混沌时间序列预测[J].物理学报，2007，56(2)：713-719.Zhang J F,Hu S S.Chaotic time series prediction based on RBF neural networks with a new clustering algorithm[J]. Physics Journal,2007,56(2):713-719.
［12］孙丹，万里明，孙延风，等.一种改进的RBF神经网络混合学习算法[J].吉林大学学报，2010，48(5) ：817-822.Sun D,Wan L M,Sun Y F,et al. An improved hybrid learning algorithm for RBF neural network[J].Journal of Jilin University,2010,48(5):817-822.

服务与反馈：

【文章下载】【加入收藏】