Open AccessExperimental investigation and ANN prediction for part quality improvement of fused deposition modeling parts
Author(s) -
Amanuel Diriba Tura,
Hana Beyene Mamo,
Yohanis Dabesa Jelila,
Hirpa G. Lemu
Publication year - 2021
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/1201/1/012031
Subject(s) - fused deposition modeling , taguchi methods , surface roughness , artificial neural network , orthogonal array , acrylonitrile butadiene styrene , deposition (geology) , design of experiments , surface finish , engineering drawing , mechanical engineering , computer science , materials science , 3d printing , mathematics , statistics , engineering , artificial intelligence , composite material , paleontology , sediment , biology
Fused deposition modeling (FDM) is the most prevalent thermoplastic additive manufacturing technology. Many input parameters and their settings have a significant impact on the quality and functionality of FDM parts produced. To enhance the quality of parts, it is critical to be able to predict surface roughness distribution in advance. The development of artificial neural network (ANN) models to forecast the impact of main FDM process factors on the part quality in terms of surface roughness while utilizing ABS (Acrylonitrile butadiene styrene) material is described in this work. Taguchi L9 orthogonal array was used to plan the experiments. Different printing input parameters such as layer thickness, orientation angle, and infill angle are used in the experiments. In terms of controllable input parameters, ANN is used to construct a predictive mathematical model. The effects of various printing settings on surface roughness were investigated using analysis of variance (ANOVA), main effect plots, and contour plots. Experiment findings and regression value are used to validate the models. The model has shown to be capable of adequately predicting responses within a maximum percentage error of 4.664 percent of arithmetic roughness average (Ra), which is a good agreement.