Uso de aprendizado de máquinas para estimar a resistência à tração de peças fabricadas por Impressão 3D

Abstract

Today’s society lives in a technological universe with several automated systems and a constant search for improvements in the efficiency and quality of processes. Due to the numerous techno- logical advances, industry 4.0 was emerged. The 3D printing is a standout tool that can be used in this era of the fourth industrial revolution and is defined as the manufacturing area responsible for building parts, starting from a virtual model, by layer by layer automatic deposition method. In this bias, the 3D printing has advantages such as great flexibility in the production of parts from the digital model, enabling the fabrication of different geometries, simple or complex. In addition to using low-cost materials and generating little residue. Many technologies have gained space in today’s society, can highlight the artificial intelligence (AI), which has several applications in different areas of knowledge, and can be defined as any technology that allows a system to demonstrate human intelligence. In this context, machine learning uses artificial intelligence to develop computational techniques, aiming to build knowledge automatically. This system is responsible for making decisions based on experiences accumulated through successful solutions. Thus, this work aims to develop a model using artificial intelligence techniques, specifically, the neural network, able to predict the tensile strength in materials manufactured by FDM (Fused Deposition Modeling) type 3D printing. Starting from the collection and construction of a data- base on three-dimensional impressions, algorithms were developed in the Python programming language, combining artificial neural networks and genetic algorithms, in a neuroevolutionary strategy, to meet the proposed objective. After analyzing the data and graphics generated by the execution of the algorithms, a model with a determination coefficient superior to 90% was elaborated, resulting in a high rate of assertiveness.