Utilizing Machine Learning to Predict the Dimensional Variation of Shafts Printed using Fused Deposition Modeling

Abstract

With the onset of the fourth industrial paradigm, additive manufacturing techniques are coming to the forefront in mechanical engineering domain. The technological burgeoning of additive manufacturing, particularly 3-D printing, has observed substantial growth in rapid prototyping, functional part manufacturing, and tooling because it has significantly reduced the manufacturing costs and processing time. One of the most commonly used techniques of additive manufacturing is Fused Deposition Modelling (FDM), examining and controlling the quality of the printed product still has massive scope for improvement. The ability to predict the tolerances and allowances in the finished product is still considered nascent. Predicting the dimensional variation of the print will enable users and engineers to get a better fit while assembling the parts, a better quality of the end product as well as better cost and time effectiveness in the complete printing process. A basic machine learning model was developed as a part of this project for predicting the variation in the outer diameter and inner diameter of printed shaft specimen which are caused by altering the significant print parameters available in the FDM printing technique. For this purpose, Polylactic Acid (PLA) as a printing material was considered, owing to its ease of printing characteristics under varying conditions.