基于神经网络与食物链算法的工艺参数的优化  

Optimization of welding process parameters based on the neural network and food chain algorithm

作　　者：谢玉婷[1] 张华[1] 鄢威[1] 张辉[1] 

机构地区：[1]武汉科技大学机械自动化学院,湖北武汉430081

出　　处：《电焊机》2015年第8期189-194,共6页Electric Welding Machine

基　　金：国家863计划项目(2012AA040101);国家自然科学基金项目(51275365)

摘　　要：为了减少堆焊过程中的能源消耗和提高批量加工的焊件合格率,建立了一种批量加工的堆焊工艺参数优化方法。首先选择焊条直径、焊接电流、电弧电压、焊接速度、焊接层数作为待优化工艺参数,焊接过程中的能源消耗和焊件合格率作为优化目标,并设置约束条件,在试验中获得优化目标与待优化工艺参数之间的离散关系并建立"黑箱"函数模型,然后用神经网络的隐式来表达输入、输出的函数关系。并利用食物链算法对形式未知的函数求优化解,求得的最优解所对应的工艺参数即为一定工艺条件下的最优工艺参数。在此堆焊工艺参数下进行批量生产可以减少能源消耗并提高产品合格率。To reduce the energy consumption and improve the weldment pass rate of welding batch processing,set up a welding process parameter optimization method for batch processing. First select the electrode diameter,the welding current,arc voltage,welding speed,the number of layers as welding process parameters to be optimized,choice the energy consumption of welding process and the weldment pass rate as the optimization goal,and set the constraints.In the tests to obtain discrete relationship between the optimization goal and process parameters to be optimized and establish a ＂black box＂ function model,and then use the implicit function neural network to express the relationship between input and output. And use the food chain to find the optimal solution of the function algorithm,corresponding to the optimal solution obtained is the optimal process parameters for the process parameters under certain conditions.In the welding process parameters under batch production can reduce energy consumption and improve the pass rate.

关 键 词：批量 合格率 能耗 神经网络 食物链 

分 类 号：TG409[金属学及工艺—焊接]