| The welding process of aluminum keyhole PAW in vertical position is a complicated and highly nonlinear process and there are many uncertain factors in it,The weld quality is affected by many technologic parameters,so it is difficult to be determined the near-optimal settings of the welding process parameters in production.The welding workers in our country make a definite study on this technology,who found out the stable existence of melting pool ensures good weld formation and weld quality.The melting pool of PAW is affected by many factors,such as the welding current,welding voltage,welding speed,distance of nozzle and weldment,flux of ion gas etc.In welding production,how to optimize these parameters settings is a problem to be solved imperatively.Based on MATLAB6.1 neural network toolbox, a BP neural network modal about vertical position PAW input-output is established.According to each value of input nodes,the output can be predicted by testing this network.When given each welding parameters of input nodes,the parameters of welding formation can be predicted. Experiments are made with the combination of each parameters.the results show that the error between parameters of welding formation and its predicting results is within 8 percent. The results of simulation experiments show that this way is practical.Simple genetic algorithm(SGA) was designed by using C++ for aluminum keyhole PAW in vertical position,which is used to decide near-optimal settings of the welding process parameters.The program about SGA shows a advantage of object-oriented design,which has a character of resuing code and provide a foundation for further design.In this paper,the near-optimal settings of the welding process parameters in keyhole PAW were determined with relatively few experiments.The study result shows that determining the near-optimal process parameters of aluminum keyhole plasma welding in vertical position by simple genetic algorithm is an efficient way.
|
PDF Full Text Request