Modal Characteristics Prediction of Robotic Machining Systems Based on Deep Neural Network
LI Fagui, WANG Ruoqi, SUN Yuwen
Key Laboratory for Precision and Non-Traditional Machining Technology of the Ministry of Education, Dalian University of Technology, Dalian 116024, China
Due to large working space and strong flexibility, serial industrial robots are widely used in the machining of large structural parts such as aircraft skin and aviation transparent part. However, the low stiffness of industrial robots and large differences in the spatial distribution of dynamic characteristics lead to low limits of their milling stability, significant variations in milling performance in different machining regions, and narrow windows of available process parameters. It is important to study the dynamic characteristics of the robot milling system during machining and to establish a positional correlation modal prediction model to improve the robot machining performance. In this paper, a modal prediction method based on deep neural network is proposed for an ABB robotic machining system. Firstly, the modal experiment of the robot processing system is carried out by using the Doppler vibrometer, and the spatial variation of each order modal is analyzed. Then, according to the actual working space of the robot, an experiment is designed to obtain the frequency response function set related to the pose, and the related modal parameters are accurately identified by the rational polynomial method. On this basis, the hyperparamter optimization method is used to establish a deep neural network prediction model, which can accurately predict the pose-dependent modal parameters in the robot workspace. Finally, the experimental results show that the prediction accuracy of this method can reach more than 80%.
李法贵,王若奇,孙玉文. 基于深度神经网络的机器人加工系统模态特性预测[J]. 航空制造技术, 2023, 66(3): 85-92,124.
LI Fagui, WANG Ruoqi, SUN Yuwen. Modal Characteristics Prediction of Robotic Machining Systems Based on Deep Neural Network[J]. Aeronautical Manufacturing Technology, 2023, 66(3): 85-92,124.