Research On Rough Set Theory & Application In Virtual Assembly System

In the background of manufacture enterprise informationization and information system intellectualization, the development status and lack of knowledge problems of virtual assembly system in manufacture enterprise are analyzed completely. The solution idea is presented by the introduction of rough set theory -one of the artificial intelligence theories- to support manufacture knowledge discovery. On this basis that several key computation optimization questions are researched completely, a new knowledge discovery model-IMSRS (Improved-MDV-SOM-RoughSet) model is presented firstly. And then the IMSRS model is employed to discover much new and useful manufacture knowledge about assembly sequence, assembly operation time distribution, and virtual device fault diagnosis.In the rough set theory research, based the factual research project, the attribute reduct normal forms are fully researched. Four research achievements are obtained: the method of CAWD(Continuous Attribute Whole Distance, CAWD) to decrease universe space dimensions, the model of IMDV(Improved Maximum Discernibility Value, IMDV) to discretize the continuous attributes, the model of Petri-RS to map and transfer the Petri Net infortmation into rough set decision information system in the form of two-dimensions table, and the rough set knowledge discovery model respectively. All these research achievements offer the more perfect theory tool base for their application for the factual questions in virtual assembly system.In the virtual assembly system research, in the background of factual project-Virtual Press Line system, the user requirements are analyzed completely, then the outline and detailed design work is done. And the developed software is validated on the current enterprise software platform economically. All this design and development work brings into the application basis and also the factual question objects for the rough set research achievements.Based on the full analysis of the assembly knowledge discovery questions, to discover the assembly sequence knowledge, the IMSRS model is applied into the two-dimension decision information tables which are transformed form Petri Net describing the virtual assembly process directly through the Petri-RS transform model. The satisfied assembly rules knowledge is obtained by the application of the IMSRS model into the question of assembly operation time distribution, which proves the correctness and effectiveness of this model. As to the virtual device fault diagnosis, the PSVM (Principal components analysis and Support Vector Machine, PSVM) model and also the IMSRS model are used to discover the diagnosis knowledge respectively, which reflects the good performance of the rough set model.In the last part of this dissertation, the summary is completed for the whole research and application work. And also the further research directions and contents for the application promotion are previewed.