Welding Path Planning of Body-in-White Based on Deep Learning
Aiming at the problem of too many empirical rules and complex paths in the welding path planning of the body-in-white,thispaperproposes an end-to-end sequence-to-sequence deep learning framework to improve the efficiency and quality of path planning by learning a large amount of historical data.A deep learning model is built based on LSTM encoder,LSTM decoder and attention mechanism.The weldpoint data is normalized and pre-sorted,and a weld point table is built to improve the training efficiency and prediction accuracy of the model.The target weld point sequence is obtained from unordered weld point sequence.The experimental results on the welding production line of the left front door of the body-in-white show that the proposed method is more accurate when facing various constraints,and the welding path planning results are more reasonable.
NETL
NSTL
万方数据
