Investigation of dimensional accuracy of material extrusion build parts using mathematical modelling and artificial neural network

Dimensional accuracy of fabricated parts made through material extrusion process is an important parameter to decide the part’s quality. Since a 3D model part is produced in layered form, the deposited layers are subjected to heat for multiple times. Also, deposited layers form bonds with adjacent layers and roads. It leads to shrinkage and distortion in fabricated parts. Process variables are also significant parameters to decide the final part dimension. Accuracy of the parts can be improved if the dimensions are predicted in an earlier stage. So, for the prediction of accurate result various mathematical models have been formulated by the researchers. Use of soft computing techniques can be one method which may also be used for prediction. Since the experiments are performed at various combination of process variables. RSM uses different mathematical models for each set of experiment, but ANN can be use at same parameters. Thus, in this paper ANN model is compared with the developed models of the selected existing literatures. Also, these models are used to find and compare the effect of process variables on dimensional accuracy. The results show that ANN model predicts the results with very less error in comparison of existing models.