Publicação
Contribution to automatic synthesis of formal theories of production systems and virtual enterprises
| Resumo: | Formal Theories (FTs) have proved their utility in “traditional” engineering areas. They permit analytical derivation of the specified system’s performance even before the system’s actual physical implementation. This is of great help giving the system’s designer and user to critical insight into the system being implemented, but before actually investing resources. However, concerning the “state-of-the-art” of the development of a FT of production systems (PS) or Virtual Enterprise (VE), we are not aware of any consistent and rigorous approach in this direction, except the initial results managed in the Centre of Production and Systems Engineering at the University of Minho. It is commonly perceived that the FTs are difficult to understand (from the cognitive point of view), difficult to learn, and therefore, difficult to develop. In terms of complexity theory, the formal theory development process is considered as a highly complex problem. Also, it is difficult to cover all the user requirements while developing a FT. On the other hand, the lack of a Formal Theory of production systems and VE is a serious obstacle to effective and efficient development, and application, especially of advanced and emerging production systems and VE concepts. In the modern times, these are the great drivers of economy and encourage innovation as well as entrepreneurship. A FT also permits an analytical synthesis, i.e. design, of the system through formal methods. This guarantees the fulfilment of the system design objectives. These are the motivational factors behind the presented thesis. The principal objective of the presented thesis is the validation of the scientific thesis concerning the problem of automatic synthesis and use of FT of production systems and VE. For the automatic synthesis of FT, the inductive inference approach is selected because of its ability to “learn” effectively from a set of examples (comparable to real-life case studies or hypothetical/abstract models for the purpose of learning). The Formal Theories are modelled through Formal Grammar and thus the problem of automatic synthesis of FT is actually reduced to the problem of Grammatical Inference. There are several methods for Grammatical Inference for different classes of formal grammars. For the presented work, the Context-free class of grammar was selected to model the intended formal theories of production systems and virtual enterprises. Through a machine learning algorithm, the formal theories were “learned” and these learned (synthesised) formal theories could generate other more complex architectures than the ones used for learning. Several experiments were carried out and a quantitative as well as qualitative analysis was carried out. The analysis of experiments’ results showed that the approach proved to be effective and with potential for real-life applications for synthesis of formal theories for virtual enterprise and similar architectures. |
|---|---|
| Autores principais: | Shah, Vaibhav Hemantkumar |
| Assunto: | Formal theories Formal grammars Automatic synthesis of formal theories Inductive inference Machine learning Manufacturing systems Virtual enterprises Teorias formais Gramáticas formais Síntese automática de teorias formais Inferência indutiva Machine learning Sistemas de produção Empresas virtuais |
| Ano: | 2015 |
| País: | Portugal |
| Tipo de documento: | tese de doutoramento |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | Formal Theories (FTs) have proved their utility in “traditional” engineering areas. They permit analytical derivation of the specified system’s performance even before the system’s actual physical implementation. This is of great help giving the system’s designer and user to critical insight into the system being implemented, but before actually investing resources. However, concerning the “state-of-the-art” of the development of a FT of production systems (PS) or Virtual Enterprise (VE), we are not aware of any consistent and rigorous approach in this direction, except the initial results managed in the Centre of Production and Systems Engineering at the University of Minho. It is commonly perceived that the FTs are difficult to understand (from the cognitive point of view), difficult to learn, and therefore, difficult to develop. In terms of complexity theory, the formal theory development process is considered as a highly complex problem. Also, it is difficult to cover all the user requirements while developing a FT. On the other hand, the lack of a Formal Theory of production systems and VE is a serious obstacle to effective and efficient development, and application, especially of advanced and emerging production systems and VE concepts. In the modern times, these are the great drivers of economy and encourage innovation as well as entrepreneurship. A FT also permits an analytical synthesis, i.e. design, of the system through formal methods. This guarantees the fulfilment of the system design objectives. These are the motivational factors behind the presented thesis. The principal objective of the presented thesis is the validation of the scientific thesis concerning the problem of automatic synthesis and use of FT of production systems and VE. For the automatic synthesis of FT, the inductive inference approach is selected because of its ability to “learn” effectively from a set of examples (comparable to real-life case studies or hypothetical/abstract models for the purpose of learning). The Formal Theories are modelled through Formal Grammar and thus the problem of automatic synthesis of FT is actually reduced to the problem of Grammatical Inference. There are several methods for Grammatical Inference for different classes of formal grammars. For the presented work, the Context-free class of grammar was selected to model the intended formal theories of production systems and virtual enterprises. Through a machine learning algorithm, the formal theories were “learned” and these learned (synthesised) formal theories could generate other more complex architectures than the ones used for learning. Several experiments were carried out and a quantitative as well as qualitative analysis was carried out. The analysis of experiments’ results showed that the approach proved to be effective and with potential for real-life applications for synthesis of formal theories for virtual enterprise and similar architectures. |
|---|