超声波喷丸成形工艺参数优化及弧高值预测

超声波喷丸成形弧高值是多个工艺参数共同作用的结果,成形工艺参数的选择及对弧高值的准确预测成为难点.本文结合正交试验法和有限元分析软件ABAQUS对不同超声波喷丸工艺参数条件下的喷丸成形过程进行数值模拟分析,研究撞针速度、撞针直径、成形轨迹间矩、喷丸区域宽度对带筋板喷丸成形弧高值的影响.对试验结果进行极差分析,探讨了喷丸工艺参数对喷丸成形弧高值的影响程度,得到较优的超声波喷丸成形工艺参数组合方案.利用正交试验得到的数据作为神经网络的训练样本,建立输入为带筋板超声波喷丸成形工艺参数,输出为成形弧高值的BP人工神经网络模型,对喷丸成形弧高值进行预测.通过样本检验该BP网络模型的准确性,实验结果数据与预测数据之间的最大误差为4.69%,从而BP神经网络能够有效代替数值模拟方法预测其弧高值,缩短工艺设计时间,提高设计效率.

Optimal analysis of ultrasonic shot peening forming(USPF) processing parameters on the formed arc height and the prediction of formed arc height

Ultrasonic shot peening forming (USPF) arc height is a result of interaction of multiple parameters. The selection of forming process parameters and accurate prediction of the arc height value are difficult issues. In the present work, the orthogonal tests and the finite element analysis, based on software ABAQUS, are carried out to study the ultrasonic shot peening process at different processing parameters. The effect of pin velocity, pin diameter, distance of forming trajectory and the width of peening strip corresponding to ultrasonic shot peening forming (USPF) on the formed arc height of stiffened plate by is studied. By the range analysis of testing results, the influence degree of USPF processing parameters can be determined and the optimal combination of peening processing parameters can be obtained. Using the data obtained from orthogonal tests as the training sample of neural network, a BP network model in which the input was the ultrasonic shot peening technological parameters and the output was the formed arc height is established to predict the formed arc height. The accuracy of the BP network was proved by the sample, the maximum error between the simulated data and predicted data is 4.69%. Therefore, BP neural network can effectively replace the numerical simulation method to predict the radius of curvature and reduce the time of process design, and improve the design efficiency.