Inteligência artificial aplicada à preparação e aplicação de insertos de metal duro

• Ano de defesa: 2024
• Autor(a) principal: Sinzato, Mateus Keniti Nakashima
• Orientador(a): Fontes, João Vitor de Carvalho
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Engenharia Mecânica - PPGEMec
• Palavras-chave em Português: Usinagem; Metal duro; Metodologia de taguchi; Preparação de aresta
• Palavras-chave em Inglês: Machining; Carbide; Taguchi methodology; Edge preparation
• Área do conhecimento CNPq: ENGENHARIAS::ENGENHARIA MECANICA::PROCESSOS DE FABRICACAO
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/21207

Grinding is an abrasive process often employed in finishing operations due to the possibility of obtaining narrow dimensional and geometric tolerances, besides low surface roughness. Besides hardened steels, that usually undergo finishing operations with grinding wheels, other materials may also be ground, like cermets: composites with ceramic particles and metallic matrix. Popularly known as hard metals, these materials mainly constituted by tungsten carbides and cobalt have wide application as machining tools. In this case, grinding is one among several abrasive processes that can be applied to the preparation of cutting edges; but it is verified a lack of publications on this purpose when related to the artificial intelligence, although their increasing number in the last few years. This work proposes to apply artificial intelligence (AI) algorithms as an optimization tool for carbide inserts grinding considering cutting speed, feed rate and radial depth of cut as input factors and geometric features of the products, such as surface roughness (Ra e Rz), as response outputs. Besides, as a complementar study, the IA application on the optimization of the straight turning was studied, using tools with two-chamfer cutting edges produced by grinding, which were evaluated the feed rate, tool nose radius and form factor (K), and their influences over the roughness Ra e Rz and residual stress generated on the 4142 alloy steel. A preliminary model was developed using the Taguchi methodology in addition to the analysis of variance (ANOVA) to complement the methodology; afterwards, it was developed optimization models using artificial Neuro-Fuzzy inference system (ANFIS) and artificial neural networks (ANN). The results demonstrated that the ANN model presented better prediction capability compared to the ANFIS model, obtaining higher capacities of becoming an optimization tool.