Research On Identification And Welding Path Planning Of Multi Type Solder Joints Of Automobile Door Panel

The manufacturing industry is the foundation of our country.With the development of society,the labor cost increases gradually.In order to improve the production efficiency and quality and reduce the manufacturing cost,it is an effective method to replace manual labor with robot to realize the full automation and intellectualization of the factory.In this paper,the intelligent recognition algorithm of solder joint and the intelligent planning algorithm of welding path are used to realize the intelligent welding of automobile door panel,so as to improve the welding efficiency of automobile door panel.In the whole welding process,solder joint identification and welding path planning are two key steps to realize intelligent welding,and also the basis to improve efficiency.This paper mainly studies the location and type of solder joint based on deep learning algorithm and the path of solder joint based on swarm intelligence algorithm.In the problem of solder joint recognition,in order to meet the needs of industry,such as fast,accurate,strong adaptability and small and many solder joints with round holes,this paper selects the algorithm of YOLOv3 based on convolution network to identify the solder joints,and through the analysis of the characteristics of the door panel solder joints,an improved YOLOv3 algorithm is proposed.In order to increase the weight of shape features and increase the recognition accuracy of solder joints,the improved algorithm mainly fuses the features of original scale 26 * 26 and original scale 52 * 52 in the original YOLOv3 network structure after down sampling the shallow features.The simulation results show that the accuracy of each type of solder joint is more than 80%,and it is improved compared with the original YOLOv3 algorithm,which verifies the feasibility and practicability of the algorithm.The welding spot position identified by the improved YOLOv3 algorithm is recorded and converted into the actual position coordinates of the door panel through a certain proportion,so as to carry out the welding path planning.In the problem of welding path planning,according to the advantages and disadvantages of the classical swarm bionic algorithm,a new improved ant colony hybrid algorithm is designed in this paper.It is based on ant colony algorithm.It takes advantage of the fast convergence speed of genetic algorithm in the early stage,uses its initial solution as the initial pheromone of ant colony algorithm to improve the initial search speed of ant colony algorithm,and uses the simulated annealing method to judge whether to update the new solution with probability p,which is easy to jump out of the local optimal solution to update the local pheromone,so as to avoid the ant colony algorithm falling into the local optimum.Through comparison and simulation,the optimal welding path planning trajectory is obtained,which proves the feasibility and the advanced nature of the algorithm.Secondly,the multi robot cooperative welding planning problem is transformed into the joint assignment problem with constraints and the single robot welding problem.The algorithm k-means is used to design and implement the solder joint assignment scheme,and the solder joint assignment schemes of 2,3 and 4 robots are obtained respectively.Taking the three robots as an example,combined with the Ant Colony Hybrid Algorithm of single robot path welding planning,the optimal path of multi robot welding path is obtained.