摘要
CO_2焊在熔滴短路过渡过程中熔池附近区域的金属飞溅、光照强度变化剧烈,严重影响焊缝特征提取的实时性和可靠性。采用正交试验法和改进的BP神经网络,建立了焊接工艺参数与CO_2焊熔池附近区域图像灰度值的映射关系。结果表明:BP神经网络模型的训练结果与试验结果的误差很小,满足精度要求。该模型能很好地反映熔池附近区域图像灰度值与焊接工艺参数的关系。
The metal spatter and light intensity at the area near the weld pool during droplet transfer process in CO2 welding change dramatically, which seriously affects the real-time performance and reliability of weld feature extraction. The mapping relationship between the welding process parameters and the image gray value of the area near the weld pool in CO2 welding was established by using the orthogonal test method and the improved BP neural network. The results show that the error between the training results of BP neural network model and the test results is very small, which meets the accuracy requirements. The model can reflect the relationship between the image gray value of the area near the weld pool and the welding parameters.
出处
《热加工工艺》
CSCD
北大核心
2017年第13期221-224,共4页
Hot Working Technology
基金
广西自然科学基金资助项目(2014GXNSFAA118310)
关键词
CO2焊
附加动量法
自适应学习速率法
BP网络
CO2 welding
additional momentum algorithm
adaptive learning rate method
BP network
