基于误差补偿的船舶双焊接机器人协同免疫网络控制  

Collaborative Immune Network Control for Marine Dual Welding Robots Based on Error Compensation

作　　者：董莹莹 申燚[1] 孙宏伟[2] 王舜 袁明新[1] DONG Yingying;SHEN Yi;SUN Hongwei;WANG Shun;YUAN Mingxin(School of Mechatronics and Power Engineering,Jiangsu University of Science and Technology,Zhangjiagang 215600,Jiangsu,China;Jiangsu Automation Research Institute,Lianyungang 222002,Jiangsu,China)

机构地区：[1]江苏科技大学机电与动力工程学院,江苏张家港215600 [2]江苏自动化研究所,江苏连云港222002

出　　处：《船舶工程》2024年第2期138-145,164,共9页Ship Engineering

基　　金：工信部高技术船舶项目(【2019】360);张家港市科技计划项目——校企合作产学研前瞻性项目(ZKYY2233)。

摘　　要：为了实现船舶双焊接机器人的高精度协同控制,提出了一种基于误差补偿的协同免疫网络控制算法。在双焊接机器人运动学分析基础上,将协同控制转化为二次型优化问题;将目标函数和关节角分别定义为抗原和抗体,借鉴免疫信息处理机制构建了包括输入层、跟踪隐层、补偿隐层和输出层的协同免疫网络;将由网络跟踪隐层评价所得递归误差经补偿隐层激活产生补偿项,再通过网络递归和疫苗接种实现船舶双焊接机器人的协同免疫网络控制器设计。4种不同轨迹的双机器人协同跟踪数值仿真表明,相比于MGNN和PDNN算法,协同免疫网络控制算法的平均跟踪误差降低19.05%,平均协同误差降低17.95%,平均关节偏移降低27.73%,验证了该算法的高精度、高协同性和高重复性控制性能。In order to achieve high-precision collaborative control of marine dual welding robots,the a collaborative immune network control algorithm based on error compensation(CINCABEC)is proposed.The collaborative control of the marine dual welding robots is transformed into a quadratic optimization problem based on kinematic analysis.The objective function and joint angle are defined as the antigen and antibody respectively,and a collaborative immune network including input layer,tracking hidden layer,compensation hidden layer and output layer is constructed by immune information processing mechanisms for reference.The recursive error obtained from the evaluation of the network tracking hidden layer is activated by the compensation hidden layer to generate a compensation term,and then the collaborative immune network controller design of the marine dual welding robots is implemented through network recursion and vaccination.The numerical simulation of collaborative tracking of two robots with four different trajectories shows that the average tracking error of the cooperative immune network control algorithm is reduced by 19.05%,the average cooperative error is reduced by 17.95%and the average joint offset is reduced by 27.73%compared with the MGNN and PDNN algorithms,which verifies the control performance of the algorithm with high precision,high cooperation and high repeatability.

关 键 词：误差补偿 双焊接机器人 疫苗 免疫神经网络 协同控制 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]