A carregar...
Publicação
Estudo e Desenho de Abordagem Auto-organizada para Sistemas de Controlo Distribuídos
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 5.25 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Com os avanços tecnológicos nos últimos anos e com o aparecimento da Internet, a indús-
tria teve de se revolucionar e adaptar a um novo mundo com um mercado extremamente
volátil e competitivo, dando origem ao que conhecemos hoje como indústria 4.0. Com a
Indústria 4.0 algumas tecnologias mais conhecidas como, por exemplo, a computação em
nuvem, Big Data e inteligência artificial, ganharam importância. No meio destas emer-
giram os sistemas ciber-físicos, tecnologia que proporciona aos sistemas físicos serem
monitorizados, controlados e integrados através de meios computacionais, que serve de
base a muitas das tecnologias que encontramos na indústria 4.0.
Para implementar esta tecnologia no mundo industrial, muitos investigadores e pro-
jetos têm-se baseado numa tecnologia proveniente da Inteligência Artificial, os Agentes.
Na indústria são utilizados sistemas multi-agente. Estes sistemas frequentemente pos-
suem capacidades de auto-organização, que se devem à aplicação de diversos algoritmos.
A maioria das arquiteturas que encontramos implementadas para sistemas distribuídos
auto-organizados, utilizam um reduzido número de algoritmos e são construídas em
torno desses algoritmos, sendo extremamente intolerantes à flexibilidade de utilizar ou-
tros algoritmos.
Neste trabalho propomos uma arquitetura para um sistema de controlo distribuído
auto-organizado para manufactura, com a capacidade de se conseguirem integrar vários
algoritmos diferentes. Esta arquitetura é baseada em agentes inteligentes e está dividida
em três camadas responsáveis por: controlo e reprodução do chão da fábrica; armazena-
mento do estado do chão da fábrica e trajetos de produtos; optimização.
A abordagem generaliza os problemas que se encontram em sistemas industriais de
produção e tenta apresentar uma base para servir de guia na construção de um sistema
ciber-físico de produção que requeira flexibilidade nos algoritmos de optimização. Esta
abordagem apresenta robustez, modularidade, flexibilidade e revela capacidades de auto-
organização.
De forma a comprovar o comportamento e as capacidades desta arquitetura, foram
realizados alguns testes através da simulação de um chão de uma fábrica e posteriormente
tiradas as conclusões sobre os resultados obtidos.
With technological advances in recent years and the emergence of the Internet, the in- dustry had to revolutionise and adapt to a new world with an extremely volatile and competitive market, giving rise to what we know today as Industry 4.0. With Industry 4.0 some better-known technologies, such as cloud computing, Big Data and artificial in- telligence, gained importance. In the midst of these, the cyber-physical systems emerged, technology that provides physical systems to be monitored, controlled and integrated through computational means, which is the foundation for many of the technologies found in Industry 4.0. To implement this technology in the industrial world, many researchers and projects have relied on a technology derived from Artificial Intelligence, the Agents. In industry, several agents are generally used, leading to multi-agent systems. These systems often have self-organisation capabilities, which are due to the application of man-made algo- rithms. Most architectures we have found implemented for self-organised distributed systems use a small number of algorithms and are built around these algorithms, being extremely intolerant to the flexibility of using other algorithms. In this work we propose an architecture for a distributed self-organised control sys- tem for manufacturing, with the ability to integrate several different algorithms. This architecture is based on intelligent agents and is divided in three layers responsible for: shop floor control and reproduction to virtual world; storage status of the shop floor and product paths; optimisation. The approach generalises the problems encountered in industrial production systems and attempts to present a basis to serve as a guide in the construction of a cyber-physical production system that requires flexibility in the optimisation algorithms. This approach presents robustness, modularity, flexibility and reveals self-organisation capabilities. In order to prove the behaviour and capabilities of this architecture, some tests were performed through the simulation of a factory floor and then conclusions were drawn on the results obtained.
With technological advances in recent years and the emergence of the Internet, the in- dustry had to revolutionise and adapt to a new world with an extremely volatile and competitive market, giving rise to what we know today as Industry 4.0. With Industry 4.0 some better-known technologies, such as cloud computing, Big Data and artificial in- telligence, gained importance. In the midst of these, the cyber-physical systems emerged, technology that provides physical systems to be monitored, controlled and integrated through computational means, which is the foundation for many of the technologies found in Industry 4.0. To implement this technology in the industrial world, many researchers and projects have relied on a technology derived from Artificial Intelligence, the Agents. In industry, several agents are generally used, leading to multi-agent systems. These systems often have self-organisation capabilities, which are due to the application of man-made algo- rithms. Most architectures we have found implemented for self-organised distributed systems use a small number of algorithms and are built around these algorithms, being extremely intolerant to the flexibility of using other algorithms. In this work we propose an architecture for a distributed self-organised control sys- tem for manufacturing, with the ability to integrate several different algorithms. This architecture is based on intelligent agents and is divided in three layers responsible for: shop floor control and reproduction to virtual world; storage status of the shop floor and product paths; optimisation. The approach generalises the problems encountered in industrial production systems and attempts to present a basis to serve as a guide in the construction of a cyber-physical production system that requires flexibility in the optimisation algorithms. This approach presents robustness, modularity, flexibility and reveals self-organisation capabilities. In order to prove the behaviour and capabilities of this architecture, some tests were performed through the simulation of a factory floor and then conclusions were drawn on the results obtained.
Descrição
Palavras-chave
Sistema Ciber-Físico de Produção Sistema Multi-Agente Auto- Organização Algoritmos Optimização