基于神经网络遗传算法的激光熔钎焊参数优化  

Laser welding-brazing parameters optimization based on BP neural networks genetic algorithm

作　　者：刘皋[1] 张朝阳[1] 黄磊[1] 姜雨佳[1] 聂昕[1] 陆海强 庄鸿武 

机构地区：[1]江苏大学机械工程学院激光技术研究所,江苏镇江212013

出　　处：《光学技术》2016年第5期431-434,440,共5页Optical Technique

摘　　要：针对激光钎焊熔深难以控制的问题,选取激光焊接功率、焊接速度等激光焊接过程中所涉及的控制参数建立基于BP人工神经网络的激光钎焊模型。根据激光钎焊模拟实验的历史数据对焊接熔深进行预测,采用遗传算法对控制参数进行优化,得到目标焊接熔深。运用MATLAB软件建立了针对铝合金/镀锌钢的激光熔钎焊过程的参数的BP人工神经网络模型,并利用遗传算法的并行和群体搜索策略,对其控制参数进行了优化,使得焊接熔深能通过过程参数精确控制,提高了接头性能。To solve the problem of the control of welding penetration, the welding power and welding speed parame- ters are selected in the process of laser welding-brazing. The model of laser welding-brazing based on BP artificial neural network is established. According to the historical data of the simulation experiment of laser welding-brazing, the welding penetration is predicted. The welding-brazing adjustable parameters are optimized by the genetic algorithm based on the established model. The goal welding penetration is got. By the help of the software of MATLAB, the process parameters of the laser welding-brazing of aluminum alloy and zinc coated steel are utilized to establish BP artificial neural network model. The parallel and group search strategy of the genetic algorithm are used to optimize its parameters. The welding penetration can be controlled accurately by the process parameters, and the performance of the joint can be improved.

关 键 词：激光技术 激光熔钎焊 MATLAB BP人工神经网络 工艺参数 遗传算法 

分 类 号：TG456.7[金属学及工艺—焊接] TN249[电子电信—物理电子学]