Um sistema de produção flexível e inteligente para planejamento de processos e otimização de desempenho empresarial

• Ano de defesa: 2023
• Autor(a) principal: Fernandes, Ederson Carvalhar
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Curitiba; Brasil; Programa de Pós-Graduação em Engenharia Mecânica e de Materiais; UTFPR
• Palavras-chave em Português: Processos de fabricação; Planejamento da produção; Fluxo de trabalho; Processo decisório; Aprendizado de máquinas; Aplicações Web; Inteligência competitiva (Administração); Algorítmos computacionais; Inteligência artificial; Manufacturing processes; Production planning; Workflow; Decision making; Machine learning; Web applications; Business intelligence; Computer algorithms; Artificial Intelligence; CNPQ::ENGENHARIAS::ENGENHARIA MECANICA; Engenharia Mecânica
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/32044

Many companies have been dedicating considerable efforts to continuously improving manufacturing processes to ensure their competitiveness and remain in the market. Value Stream Mapping is a strategic tool that enables the visualization of the macro of production to assist in planning and decision-making. It is a process mapping that considers the workflow of a product from the arrival of raw material to the result that is delivered to the customer. Despite the benefits this tool has provided to organizations, the time for its development is still very high, often ranging from weeks to months, as it involves manual data entry. This manual process increases the likelihood of errors in the analysis due to the inherent potential for human mistakes. It is important to note that the manual development of VSM carries the risk of simple errors occurring, which can impact the accuracy and reliability of the obtained results. Using data science has brought several improvements to the methods and tools of organizations. Then, to obtain contributions to engineering, this study presents the development of a dynamic web system for current and future state process mapping using data analysis and machine learning. Through the DSR (Design Science Research) methodology that aims to develop innovative solutions for complex problems, a Docker virtualization system is built, through an ETL (Extraction, Transformation, and Loading) approach, for the management of six containers with applications that communicate with each other in an isolated environment, in order to bring together all the necessary resources for the organization, visualization, and dynamic updating of the current mapping, as well as simulations and predictions in building future mappings to aid management decision-making. As a result, the performance, security, interoperability, generality, flexibility, and efficiency of the developed system are demonstrated through its application and evaluation in a production flow of an agricultural equipment company in Curitiba, Brazil. The current state mapping was carried out with the most current data provided by the company, and four alternatives for improving the process were predicted, with the evaluation and approval of the company’s engineers. The prediction system provides greater agility and support for decision-making. Combining a method already consolidated in the manufacturing industry with digital technologies contributes to identifying new patterns, critical areas, and complex relationships between variables, providing additional insights into the production process.