摘要
使用机器人进行焊接作业是提升工业生产效率的重要手段,合理的工艺参数是保障机器人焊接质量的关键。针对现有焊接工艺参数优化方法易陷入局部最小、训练时间长、拟合精度不高的问题,提出了一种混沌麻雀搜索算法优化双权值神经网络的新算法。该方法利用混沌麻雀搜索算法的全局搜索能力为双权值神经网络的方向权值、核心权值、幅值选取最优参数。提出了基于新算法的焊接工艺参数优化方法,建立焊接工艺参数与焊接质量之间的映射模型,获取最优焊接参数。分别采用模拟数据与实测激光焊接实验数据对所提方法进行测试,结果表明,混沌麻雀搜索算法优化的双权值神经网络迭代速度快、拟合精度高,较传统的双权值神经网络和径向基神经网络性能更优,适用于工业生产中焊接机器人工艺参数的高效设定。
Robot welding is an important means to improve the efficiency of industrial production,reasonable process parameters are the key to ensure the quality of robot welding.In this paper,aiming at the problems that existing welding process parameter optimization methods tend to fall into the local minimum,long training time and low fitting accuracy,a new algorithm of chaotic sparrow search algorithm is proposed to optimize the double weight neural network.The global search ability of Chaotic sparrow search algorithm is used to select the optimal parameters for the direction weight,core weight and amplitude of the double weight neural network.The welding process parameters optimization method based on chaotic sparrow search algorithm is proposed to optimize the double weight neural network,and the mapping model between welding process parameters and welding quality is established to obtain the optimal welding parameters.The proposed method is tested by using the simulated data and the measured laser welding experiment data,respectively.The results show that the double weight neural network optimized by chaotic sparrow search algorithm has faster iteration speed and higher fitting accuracy,and has better performance than the traditional double weight neural network and radial basis neural network,which is suitable for the efficient setting of welding robot process parameters in industrial production.
作者
朱广明
华亮
赵佳皓
羌予践
ZHU Guangming;HUA Liang;ZHAO Jiahao;QIANG Yujian(School of Electrical Engineering,Nantong University,Nantong 226019,Jiangsu,China)
出处
《实验室研究与探索》
CAS
北大核心
2023年第7期48-53,共6页
Research and Exploration In Laboratory
基金
江苏省高等学校自然科学研究重大项目(19KJA350002)。
关键词
焊接机器人
双权值神经网络
混沌麻雀搜索算法
参数优化
welding robot
double weights of neural network
chaotic sparrow search algorithm
parameter optimization
