Controle da produção utilizando estratégias de consenso multiagente

• Ano de defesa: 2023
• Autor(a) principal: Sousa, Alex Luiz de
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Curitiba; Brasil; Programa de Pós-Graduação em Engenharia Elétrica e Informática Industrial; UTFPR
• Palavras-chave em Português: Sistemas flexíveis de fabricação; Sistemas multiagentes; Processos de fabricação; Sistemas de controle inteligente; Planejamento da produção; Controle de produção; Programação de produção; Flexible manufacturing systems; Multiagent systems; Manufacturing processes; Intelligent control systems; Production planning; Production control; Production scheduling; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA::ELETRONICA INDUSTRIAL, SISTEMAS E CONTROLES ELETRONICOS; CNPQ::ENGENHARIAS::ENGENHARIA DE PRODUCAO::GERENCIA DE PRODUCAO::PLANEJAMENTO, PROJETO E CONTROLE DE SISTEMAS DE PRODUCAO; Engenharia Elétrica
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/32959

Flexibility in production processes, related to the ability to reconfigure machines and the diversity of products produced, is seen by the industry as a strategy for gaining a competitive advantage. To ensure flexibility and reconfiguration, flexible manufacturing systems have been updated with concepts and technologies, representing a new production model aligned with Industry 4.0. However, designing intelligent and adaptable controls for flexible manufacturing systems is challenging. Approaches employing multi-agent systems have been well accepted for this process due to the decentralized and cooperative way agents solve optimization problems and deal with complex tasks. Multi-agent consensus, which can be seen as an algorithmic procedure that enables convergence to new information states, can be exploited to compose more efficient control solutions in flexible manufacturing systems. Controlling production in these systems requires that agents also have the skills to deal with unforeseen and undesirable situations, such as rush orders, deadlocks, and other production deadlocks. Much research proposes rescheduling approaches as a solution to rush orders and deadlocks, but they are limited to specific production environments or are difficult to adapt to other production scenarios. However, certain types of deadlock can be better resolved in advance in production planning, taking into account the type and capacity of machinery available. Machine and equipment buffer restrictions, as well as the transportation system used to move parts or products between machines, need to be taken into account when planning production. Rush orders, which have significant disruptive capacity in production time but are accepted by the industry due to the competitive market, also depend on adequate controls for the constraints of the production environment. In addition to deadlocks in the system, the predictability crucial for industries to meet their demands efficiently, at lower costs and at appropriate levels is also at risk of being compromised. The objective of this thesis is to generate new solutions for production scheduling and control using multiagent consensus strategies and the readjustment of predictive schedules. The studies developed include flexibility in production processes, decentralization of production, a trend in Industry 4.0, solutions to deadlock problems, predictability conditions with specialized scheduling, support for the insertion of rush orders in flexible manufacturing systems, and the applicability of the approaches to other production environments. Experiments have proven that multi-agent systems efficiently decentralize order-controlled flexible manufacturing systems. Readjustment filters were applied to predictive schedules to avoid potential production deadlocks, and multi-agent consensus proposals ensured agent cooperation on tasks related to normal and rush production orders. Taking advantage of consensus to insert rush orders at production time, avoiding deadlocks due to overlapping and buffer constraints, is an innovation for flexible manufacturing systems and an efficient approach to avoiding computationally onerous rescheduling methods.