基于BP神经网络的连杆衬套强力旋压轴线直线度预测  

Prediction of axial straightness of connecting rod bushing power spinning based on BP neural network

作　　者：吉梦雯 樊文欣[1] 张涛[1] 郭代峰 李兴艳 JI Meng-wen;FAN Wen-xin;ZHANG Tao;GUO Dai-feng;LI Xing-yan(College of Mechanical and Power Engineering, North University of China, Taiyuan 030051 , China)

机构地区：[1]中北大学机械与动力工程学院

出　　处：《塑性工程学报》2018年第1期137-141,共5页Journal of Plasticity Engineering

基　　金：山西省自然科学基金资助项目(2012011023-2);山西省高校高新技术产业化项目(20120021)

摘　　要：强力旋压成形工艺的轴线直线度误差对采用该工艺生产的连杆衬套的产品使用可靠性有很大影响。采用单因素试验设计方法,利用数值模拟技术手段,获得了以减薄率、进给比、首轮压下比、主轴转速为试验因素,内、外轴线直线度误差为评价指标的试验数据。基于试验因素和评价指标应用BP神经网络技术建立了4-10-2的三层神经网络结构模型,得到了减薄率、进给比、首轮压下比、主轴转速和轴线直线度误差之间的非线性关系。用试验所得到的数据对模型进行了训练和预测,并将预测值与仿真值相比较。结果表明：该模型可以有效预测连杆衬套强力旋压轴线直线度误差。The axis straightness error of power spinning process has great influence on the reliability of the connecting rod bushing products produced by the forming process. Using the single factor experimental design method and through the numerical simulation,the test data with the experiment factors of reduction rate,feed ratio,the first run down ratio,spindle speed and evaluation index of internal and external axis straightness error was obtained. Based on the test evaluation factors and evaluation index and by using BP neural network technology,the 4-10-2 three layer neural network model was established,the nonlinear relationship among the reduction rate,feed ratio,the first run down ratio,the spindle speed and the straightness error was obtained. The model was trained and predicted by the experimental data,and the predicted value was compared with the simulation value. The results show that the model can predict the axis straightness error of the connecting rod bushing power spinning effectively.

关 键 词：连杆衬套 强力旋压 轴线直线度误差 BP神经网络 

分 类 号：R146.11[医药卫生—公共卫生与预防医学] TP311[自动化与计算机技术—计算机软件与理论]