基于正交实验与BP神经网络的2AL2激光切割工艺参数优化  

Process Parameters Optimization for 2AL2 Aluminum Alloy Laser Cutting Based on Rthogonal Experiment and BP Neural Network

作　　者：伍文进[1] 徐中云[1] 滕凯[1] 严帅 张磊[1] WU Wenjin;XU Zhongyun;TENG Kai;YAN Shuai;ZHANG Lei(Jiangsu Key Laboratory of Large Engineering Equipment Detection and Control,Xuzhou Institute of Technology,Xuzhou Jiangsu 221008,China)

机构地区：[1]徐州工程学院江苏省大型工程设备装备检测与控制重点实验室,江苏徐州221008

出　　处：《机床与液压》2018年第10期13-17,共5页Machine Tool & Hydraulics

基　　金：徐州市科技计划项目(KC16GZ0i5);江苏省科技计划项目(BC20140071)

摘　　要：运用正交试验对2A12铝合金激光切割参数中激光功率、切割速度、气体压力的工艺数据进行极差、方差分析,得到综合质量最优的因素组合。并以切口粗糙度为研究对象,建立其BP神经网络预测模型,训练后的模型在验证样本测试中的预测值和实际值之间误差较小,从而证明了建立BP神经网络来预测激光切割切口表面粗糙度的可行性,在实际生产中对指导激光切割获得较好的切割质量有一定的实用价值。2AL2 Aluminum Alloy cutting experiment was carried on CO2 laser cutting machine. The process data of laser power,cutting speed and gas pressure were analyzed by means of range and variance analysis of orthogonal test,and the optimum technological conditions for the best quality were determined. Taking the incision roughness as the object of study,by building a BP neural network prediction model,the error between the actual value and prediction was small,which proved that BP neural network was feasible to predict the laser cutting surface roughness. It has certain practical value in the actual production to obtain good cutting quality in laser cutting.

关 键 词：激光切割 正交试验 极差分析 方差分析 神经网络 

分 类 号：TG485[金属学及工艺—焊接]