基于BP神经网络的强力旋压成形连杆衬套力学性能预测  

Prediction of Mechanical Property of Power Spinning Forming Connecting Rod Bushing Based on BP Neural Network

作　　者：冯志刚[1] 樊文欣[1] 赵俊生[1] 韩双凤[1] 幸华超 陈翔[2] 

机构地区：[1]中北大学机械工程与自动化学院,山西太原030051 [2]中北大学机电工程学院,山西太原030051

出　　处：《热加工工艺》2014年第5期89-91,95,共4页Hot Working Technology

基　　金：山西省自然科学基金项目(2012011023-2);山西省高校高新技术产业化项目(20120021);中北大学第十届研究生科技基金项目(20131018)

摘　　要：强力旋压成形工艺与成形件的力学性能的关系是复杂的、具有高度非线性,一般很难用公式来表达。通过正交试验,获得了实验因素(旋轮与芯模间隙值、热处理温度、进给比)下,评价指标(布氏硬度、屈服强度、抗拉强度、伸长率)的实验数据。运用BP神经网络技术建立了成形件力学性能预测模型,并用实验所得的数据对模型进行了训练。结果表明:模型具有较高的精度,能够很好地反映工艺参数与成形件力学性能之间的复杂关系。该结果能为工艺参数的优化提供理论依据。The relationships between power spinning forming process and mechanical property are complex, and it has high non-linear. In general,it is difficult to express by formula. The experimental data of the evaluation （Brinell hardness, yield strength, tensile strength, elongation） under the experimental factors （the gap between the mandrel and roller, heat treatment temperature and feed ratio） were obtained based on the orthogonal design method. The prediction model of mechanical property was established based on BP neural network, and training of the model was performed by using experimental data. The results show the model has high accuracy, which can well reflect the complex relationship between process parameters and mechanical properties. The results can provide the theoretical guidance for optimization of process parameters.

关 键 词：强力旋压 工艺参数 力学性能 BP神经网络 

分 类 号：TG376[金属学及工艺—金属压力加工]