Application of artificial neural network for predicting weld quality in laser transmission welding of thermoplastics

The present work establishes a correlation between the laser transmission welding parameters and output variables though a nonlinear model, developed by applying artificial neural network (ANN). The process parameters of the model include laser power, welding speed, stand-off distance and clamping pressure; whereas, the output parameters of the model include lap-shear strength and weld-seam width. Experimental data is used to train and test the network. The present neural network model is used to predict the experimental outcome as a function of input parameters within a specified range. Linear regression analyses are performed to compute the correlation coefficients, to measure the relationship between the actual and predicted output values, for checking the adequacy of the ANN model. Further, a sensitivity analysis is performed to determine the parametric influence on the model outputs. Finally, a comparison is made between the ANN and multiple regression models for predicting laser transmission weld quality. The same data set, which are used to develop the ANN model, are also used to develop the multiple regression models. The simulation data obtained from the neural network confirms the feasibility of this model in terms of applicability and shows better agreement with the experimental data, compared to those from the regression models.