AZ31B镁合金电流辅助旋压回弹角预测及工艺参数优化

Prediction on springback angle and process parameter optimization in electro-assisted spinning for AZ31B magnesium alloy

在金属旋压工艺中,回弹是一种无法避免的成形缺陷,为了减小旋压制件的回弹,在电流辅助旋压的基础上,以AZ31B镁合金旋压件为研究对象,通过正交实验探究了电流强度、主轴转速、旋压轮进给速率与回弹角的关系,对实验结果进行了极差分析和方差分析,得到了工艺参数对回弹角的影响规律及最小回弹角的工艺参数组合。以实验数据作为训练样本,建立了BP神经网络模型进行回弹角预测,将实验得到的工艺参数组合作为输入,进行回弹角预测及实验验证,结果表明:BP神经网络模型的预测结果与实验结果的相对误差小于3%,能够较准确地预测回弹角,为实际生产和进一步的实验研究提供了理论指导。

Springback is one of the unavoidable forming defect in metal spinning process.Therefore,in order to reduce the springback of spinning parts,on the basis of electro-assisted spinning,for AZ31B magnesium alloy spinning parts,the relationship between current intensity,spindle rotate speed,spinning roller feeding rate and springback angle was explored by orthogonal experiment,and the experiment results were analyzed by range analysis and variance analysis to obtain the influence laws of process parameters on the springback angle and the process parameters combination of the minimum springback angle.Then,taking the experimental data as training sample,the BP neural network model was established to predict the springback angle,and the combination of the process parameters obtained in the experiment was used as input to predict the springback angle and conduct the experimental verification.The results show that the relative error between the BP neural network model prediction results and the experiment results is less than 3%,which can accurately predict the springback angle.Furthermore,it provides theoretical guidance for actual production and further experimental research.