Research On History-independent Technology Of Semantic Feature Modeling

Feature modeling can provide high-level conception design and function design for designer in special domain. Parametric and variational feature modeling system is built by combining traditional modeling technique with parametric and variational technique. These systems depend on creation history of features, and the whole design process must be repeated according to the design history if the model needs to be modified in later design period. This results in the low efficiency of model design.History-independent modeling is complement to feature modeling. In history-independent modeling system, the designers only need to pay attention to design results, but not design process. The research on history-independent feature modeling and promoting flexibility and efficiency of modeling system is a new trend of modeling system, which has theoretical and practical significance. This thesis includes the following contents:Firstly, a method of naming and coding topological entities is proposed, which is based on features, and this method aims at dynamic and reversible characteristics of history-independent modeling. The semantic and attribute of feature is unchangeable in process of modeling. Topological face is named by using feature name, and topological edge and vertex are named by using adjacent faces of them. On the basis of feature name, the split sub-faces are distinguished according to the order that sub-faces are sorted on the parent face. References are used to distinguish the split edges and the multi-edges that produced by surfaces intersection. A uniform coding format is proposed, which is based on topological entities naming. The conception of virtual topological entity and sub-edge are proposed, then inherit, split and merge operation are processed which aim at characteristics of history-independent modeling. History-independent modeling operation is realized through maintaining relations between topological entities in the model. Secondly, model process and maintain methods during editing and modifying are proposed according to model operation methods. The history-independent feature model operations are summarized as translation, extension and rotation. For features or part models that are directly affected by model modification, the result model can be generated by constraint solving and to be displayed if there are no variations in topological structure. The model is processed based on topological entities coding when topological structures are changed. The split topological entities are processed according to inherit rules and split operation, and the merged topological entities are processed by utilizing sub-edges and virtual topological entities. For features or part models that are indirectly affected by model modification, the relations among these features are analyzed by making use of feature dependent graph (FDG). The dynamic feature modification precedence criteria are built and property of feature precedence is given. The appropriate result model is produced according to these criteria and property. Validity checking and recovery method during process of model modification are proposed.Thirdly, constraint solving mechanism is proposed in process of history-independent model modification. The basic definition of geometric constraint is built based on Euler parameter expressions of geometric entities. Then the algebraic equations of common project geometric constraints are derived to establish mathematic model of geometric constraints. The algebraic equations of geometric constraints are transformed into optimization problem and are solved by optimization method. An improved chaos search strategy is proposed, and which is introduced into particle swarm optimization algorithm (PSO). PSO searches solutions rapidly in early evolution period, which is used to search primary solutions. When the algorithm gets into the local extremum or finds preferable solutions, the improved chaos search strategy is used to activate the particles and to search the global best solutions accurately.At last, the update and display strategy of feature model is studied, which can satisfy history-independent modeling. The model process procedure is divided into two steps, namely pretreatment and practical solving. FDG is used to store feature information and relations of dependence and constraints between features. A method of decomposing FDG is proposed, which decomposes FDG into macro-geometry and rigid body. Macro-geometry is affected by the modified parameters, and rigid body is not affected. The macro-geometry is determined by correlation set, which is the principal part of constraint soling. The macro-geometry is expressed in geometric constraint graph. Each constraint relation can be expressed by constraint expressions and be solved in local coordinate system. After the constraints have been solved, the method of updating results in geometric constraint graph is studied. At last, the method of collaborative design based on the proposed system is studied.