Abstract: To reduce the spatial error of heavy gantry-type automated fiber placement machine and improve the quality of composite material laying and forming,the methods of spatial error modeling and parameter identification were proposed.The composition factors of the spatial error were analyzed.According to the multi-body system theory,the theoretical kinematics model and the actual kinematics model of automated fiber placement machine were established and then the spatial error model of the end of automated fiber placement machine was established.The third-order Chebyshev polynomial was used as the basis function to express the PDGE errors,and the spatial error component with undetermined error parameters was determined.The optimization objective function of the spatial error parameter identification of automated laying machine was established,and Levenberg-Marquardt (LM) algorithm was used to solve the unconstrained nonlinear least squares optimization problem,and 162 error model parameters were obtained.By measuring the diagonal line of the working space of the automated fiber placement machine and the laying path of the ARJ21 wingtip wing panel,the validity of the error modeling and parameter identification method was verified.

Key words: automated fiber placement machine, multi-body system, spatial error model, Levenberg-Marquardt algorithm, parameter identification

摘要： 为减小复合材料自动铺丝机定位误差,改善复合材料铺放成型质量,提出了大型龙门式自动铺丝机空间误差建模及参数辨识方法。首先分析了龙门自动铺丝机空间拓扑结构及几何误差组成要素,基于多体系统理论分别建立了自动铺丝机理论与实际运动学模型,进而获得了自动铺丝机末端空间误差模型;然后采用三阶切比雪夫多项式为基函数描述与位置有关的几何误差(PDGE),确定含待定误差参数的空间误差分量,建立自动铺丝机空间误差参数识别优化目标函数,采用Levenberg-Marquardt算法(LM算法)求解该无约束非线性最小二乘优化问题,获得了162项误差模型参数;最后通过测量铺丝机工作空间体对角线和ARJ21翼梢小翼壁板铺放路径,验证了误差建模及参数辨识方法的有效性。

关键词: 自动铺丝机, 多体系统理论, 空间误差模型, LM算法, 参数辨识