BP神经网络在铝合金板材激光弯曲成形中的应用

Application of BP Neural Network in Laser Bending Forming of Aluminum Alloy Sheet

在铝合金板材的激光弯曲成形中,工艺参数(激光功率、扫描速度和扫描次数)的不同组合使板材的弯曲角度也不同,而且从试验数据中很难得出激光工艺参数与弯曲角度之间的规律。本文用BP神经网络技术对铝合金板材的激光弯曲成形结果和工艺参数进行了预测。结果表明,预测值与真实值之间的误差都稳定在10%以内。精度比较高,说明运用BP神经网络对激光弯曲角度进行预测是可行的。

In laser bending forming of aluminum alloy sheet, the different combination of process parameters （laser power, scanning speed and scan times） makes bending angle oft_he sheet different, and it is difficult to get the law between the laser process parameters and bending angles from the experimental data. Laser bending forming results and process parameters of aluminum alloy sheet were predicted by using the BP neural network technology. The results show that the error between predictive value and true value is less than 10%. And the accuracy is higher, which illustrates that laser bending angle predicted by using the BP neural network is feasible.