Geometrijsko modeliranje objekata sa elementima slobodne forme podržano analizom njihovih semantičkih odlika
AuthorTrifunović, Milan B.
Committee membersManić, Miodrag
MetadataShow full item record
Product geometry is usually defined by combination of regular and free-form shapes. One of the biggest challenges associated with geometric modeling and digital reconstruction of free-forms comes from uniqueness and complexity of these shapes. While modeling or digitally reconstructing free-form shapes, the designer has to choose the right set of geometric features from the palette provided by CAD program and then compose them in a way which will enable the most accurate modeling or reconstruction of the geometry. Designer primarily relies on personal experience gained through work with objects with free-form elements of similar geometric and functional features. The main task in dissertation is to create KBE (Knowledge Based Engineering) add-in in CAD software package, by means of which the process of design or/and reverse modeling of objects with free-form elements would be (as much as possible) automated, leading to significant savings in time and costs of designing. Automatization ...of the process of design and reverse modeling of objects with free-form elements here means making decisions about which geometric features should be used and how should they be composed. By studying the work of the designer, it can be concluded that automatization of process of geometric modeling or/and digital reconstruction of objects with free-form elements requires automatization of semantic interpretation of their features. Semantic data model is necessary for modeling semantic features of geometry of objects with free-form elements. Since designer tries to solve the problem faster by applying previous solutions (or parts of solutions) for similar problems, existence of case-based / analogy-based mechanism which will operate at the semantic network level is necessary. Taking into account limitations of current KBE methods and semantic models, as well as potentials of the Active Semantic Model (ASM), there exists reasonable assumption that ASM, with necessary and appropriate upgrade, would be very good candidate for KBE add-inn for supporting process of geometric modeling or/and digital reconstruction of objects with free-form elements. For that reason, the tasks in research (dissertation) are: to develop procedure for determining topological analogy between parts of ASM semantic network, which is necessary for retrieval of similar previously considered cases, to improve context upgrading procedure, which is essentially implemented mechanism for case-base / analogy-based reasoning in ASM, and to investigate and prove applicability of upgraded ASM for automatization of the process of geometric modeling or/and digital reconstruction of objects with freeform elements. ASM was created as a kind of semantic network in which associations play the most important role. The association of ASM is a kind of semantic relation between the network nodes which is defined as a specific informatic structure characterized by the nine attributes or parameters. Semantic network associations are organized into contexts (sets of semantically close associations). Each network node represents a concept which is characterized just by its name. The main advantage of ASM, compared to other semantic models, lies in semantic network structure where semantic relations or associations (the term that is used in ASM) between the nodes of the semantic network are defined as special informatic structures characterized by specific number of attributes or parameters. Such association structure was created in order to support the thesis stating that the knowledge people have about things (visual representations, objects, situations, etc.) is contained in associations between concepts that abstractly represent those things. Procedure for determining topological analogous association plexuses (association plexus is subset of context) and contexts of the ASM semantic network was developed. This procedure enables retrieval of semantic network parts which have same type of topology and the same structure. Type of topology stands for combination of appropriate values of topological parameters (roles of concepts, type, character and direction) of associations in association plexus. In the context of graph theory this procedure can be categorized as recognizing simultaneously maximum common subgraph of input graph and each of the remaining graphs of the ASM semantic network. ASM semantic network, consisting of associations organized into contexts, can be represented as a set of labeled directed multigraphs with unique node labels. Context upgrading procedure, which represents original method of analogy-based reasoning in the semantic network, was upgraded. This procedure, together with the previous, is essential for reasoning in case of uncertainty, i.e. in situations when the input is not known in advance, like with object with free-form elements. Process of analogy-based reasoning realized in this way enables autonomous, flexible and analytical semantic interpretation of data embedded in the semantic network. Another advantage of this procedure is semiautomatic adaptation of solutions of previously considered cases. The AcSeMod application, implementing the ASM structure and accompanying cognitive data processing algorithms, was developed. This application was used for testing of the approach. Testing of the approach was conducted over the problems that arise in the process of reverse modeling of human bone geometry, concretely human femur. During the testing of the approach, procedures for reverse modeling of separate regions of the femur, isolated on the basis of anatomical and morphological features of femur, are proposed as a result. Procedures for reverse modeling of trochanteric region, femoral shaft and distal femur were proposed. Results of testing showed that the usage of semantic structures of ASM and associated cognitive data processing procedures provides the ability to perform semantic categorization and interpretation of geometric, functional and technological features of objects with free-form elements. In addition, ASM was shown to be able to autonomously create responses (procedure for reverse modeling) for the unpredicted data input (which is the situation where object with free-form elements is the considered input). Finally, ASM, improved and upgraded in this way, has shown the potential to significantly improve computer aided design and reverse modeling, at the same time actively supporting collaborative product development process.