Weld Shape Prediction Of Wire And Arc Additive Manufacturing Based On Artificial Neural Network

Additional material manufacturing technology is a method of forming a relatively short history,it is accumulated on the basis of the theory of invention comes in discrete.Artificial neural network is a neural network to animal behavior characteristics as the principle,algorithm for information processing model,it is people characteristics of brain or neural network abstract processing after.In the process of increasing material manufacturing,there is a very important work,which is based on the known welding parameters are calculated to determine the weld size,or is made according to the weld size,the welding process parameters.The author in this paper to robot surfacing weld were investigated.In order to reveal the welding speed,wire feed speed,welding voltage,nozzle height,bead width and height associated with each other.First,create a model in the study of influencing factors of weld forming,the model variables for welding speed,wire feed speed,welding voltage,voltage nozzle,nozzle height,the dependent variable is the bead height and width by two rotational regression design to find test parameters,to carry out welding experiments of monolayer,to obtain the sample data two times.General regression models to create a classic,to determine the independent variables and the dependent relationship between variables,measured by the accuracy of artificial neural network model.Then an artificial neural network model based on the positive test sample measured data,select the data within the scope of test parameters of different test parameters and form validation samples in the experimental design,the welding test,collecting test data.By forming experimental data,reveal the size error of artificial neural network model,the error of two regression model comparison.On the basis of summing up the process parameters how to influence the bead formation.Artificial neural network model is created with reverse welding parameters and weld forming size using the sample data,using data validation sample detection and reverse prediction errors with two quadratic regression general model error comparison.Through the verification calculation and error comparison of reverse error.Based on this model is created,the inverse model of closed loop fusion together the iterative feedback system.Finally,the test data were collected by the method of single pass multilayer welding,and the performance of the closed loop feedback iterative system model in the prediction of welding shaping dimensions was revealed.