摘要
借助Levenberg-Marquardt算法,建立了铅黄铜超塑性拉伸温度、初始应变速率与延伸率、流动应力之间的BP网络预测模型,分析了拉伸温度、初始应变速率与延伸率和流动应力之间的关系,得出了铅黄铜最佳的超塑性条件,并以此为依据,进行了铜合金轴承保持架的超塑性成形试验。结果表明,人工神经网络方法是优化铅黄铜轴承保持架超塑性成形工艺参数的有效方法,所预测的铅黄铜最佳超塑性条件能够满足保持架超塑成形的需要,且在最佳超塑性条件下成形保持架具有明显的经济效益。
The BP network predicting models of lead brass among superplastic tension temperature, initial strain rate and elongation, flow stress were established using the Levenberg-- Marquardt method. The relationships among superplastie tension temperature, initial strain rate and elongation, flow stress were analyzed also. Then optimal superplastic forming parameters were found. According to the parameters, the test of superplastic forming of the solid cage was performed. Results show that the artificial neural network is an effective way of optimizing the process parameters, the parameters meet well the demands of superplastic forming of the solid cage. And this forming process has obviously economic benefits under the optimal superplastic conditions.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2007年第23期2786-2788,2859,共4页
China Mechanical Engineering
基金
国家自然科学基金资助项目(50575185)
河南省高校杰出科研人才创新工程资助项目(2004KYCX020)