The aim of this thesis is to provide a support to the beginning of the conceptual development phase by offering designers the possibility to computationally synthesise technical processes in order to obtain operand transformation variants in the respect to known technological principles. To accomplish the aim the following objectives had to be met: a theoretical objective which considered the development of a method for generation of operand transformation variants; and an empirical objective as the implementation of the method as a computational tool built to a stage that allows verification of the research results. First, it was necessary to understand the phenomenon of problem solving and cognitive aspects of synthesis as a part of the problem solving activity. Then, the state-of-the-art review on the Computational Design Synthesis (CDS) [2] was conducted the purpose of which was the determination of theoretical and methodological background of the current research projects and the comparison and the systematization of those in order to focus this research. The efforts where turned to the exploration of the existent mathematical concepts which could be used for the modelling of technical processes and related synthesis method. Based on the findings from the field of CDS it was concluded to conceive the method as a knowledgebased with the solution emerging as a result of successive application of production rules in which the knowledge about technical processes and working principles is formalised. The theoretical objective concluded with the main scientific contribution of this thesis: (1) the creation of multigraph based formal model of technical process, (2) the definition of graphgrammar based formal model of technical process synthesis, (3) addition of stochastic search to technical process synthesis by applying grammatical evolution [3]. Within the empirical objective a computational tool was realised on the foundations of the developed method. During the research it was found that knowledge about technical processes still does not exists in the accessible open taxonomies or ontologies as per se, which required to propose (4) knowledge formalisation suggestions when defining the graph grammar of technical processes.
2011
PhD thesis
