摘要
如何找出特定的最佳工艺参数是焊接工作者重要而又艰巨的一项工作,是进行焊接加工时首先需要解决的问题。在全面考虑BP神经网络(Back propagation neural network)的非线性映射功能和GA(Genetic algorithm)全局寻优方法的基础上提出了综合利用回归正交表、人工神经网络(ANN)及遗传算法(GA),在所有可能的焊接工艺参数范围内自动搜寻最佳工艺参数的方法,研究中比较了不同种群大小、不同交叉概率对精度及效率的影响。结果表明,该方法具有适应性广、可靠性高的优点,由于可以大大减少试焊次数,具有良好的推广价值。
It is important and hard work to find out the best welding parameters before working at first. On the study of nonlinear mapping function of BP neural network（BP NN） and global optimizing ability of genetic algorithm （GA）, a method combined quadratic regressive orthogonal design with artificiaI neural network and genetic algorithm is introduced. The method can find out the best welding parameters within the whole range of possibilities. The different influence caused by various population size and different crossover fraction is also studied. Results testify that this method can not only reduce the testing number hut also has the virtue of accuracy, efficiency and extensive adaptability.
出处
《材料导报》
EI
CAS
CSCD
北大核心
2009年第24期69-72,共4页
Materials Reports
基金
中国博士后科学研究基金(20080441085)
江西省教育厅科技项目(GJJ08451)
关键词
焊接参数
全局寻优
BP网络
遗传算法
welding parameters, global optimization, BP neural network, genetic algorithm
