基于BP神经网络的光纤激光切割切口粗糙度预测

为了研究工艺参量对光纤激光切割切口质量的影响,进行了切割T4003不锈钢试验,分析了工艺参量与切口质量之间的关系。采用基于误差反向传播算法的人工神经网络,建立了激光功率、切割速率、辅助气体压力等工艺参量与切口粗糙度之间的预测模型。对切割试验采集的训练样本进行了网络训练,并利用测试样本对训练模型进行验证。结果表明,随着激光功率增加,切口粗糙度增大;随着切割速率和辅助气体压力增加,切口粗糙度减小。神经网络预测模型精度较高,网络训练效果良好,预测值与试验样本值间的最大相对误差为2.4%。训练后检验精度较高,检验样本最大相对误差仅为6.23%。该模型可有效预测激光切割切口表面粗糙度,同时为合理选择及优化工艺参量,提高激光切割质量提供试验依据。

Roughness prediction of kerf cut with fiber laser based on BP artificial neural networks

In order to study effects of process parameters on kerf quality of fiber laser cutting, the relationship between process parameters and kerf quality was analyzed based on the test of laser cutting T4003 stainless steel. The prediction model between the main process parameters, such as laser power, cutting speed, assistant gas pressure and kerf roughness was established based on error back propagation artificial neural network. The samples collected by the cutting test was network trained and the training model was inspected by the test samples. The results show that, kerf roughness increases while laser power increases and kerf roughness decreases while cutting speed and assist gas pressure increase. The neural network prediction model has high precision and the network training has good effect. The maximum relative error between the predictive values and the test sample value is 2.4%. After training, the prediction model has high inspection precision, the maximum relative error of the test sample is only 6.23%. The model can predict the laser cutting kerf roughness effectively and can provide the experiment basis for selecting and optimizing process parameters and improving laser cutting quality.