Research On Modeling And Dynamic Numerical Solution Of Mechanism System Virtual Prototyping Platform

As one of the important branches of computer aided engineering, mechanism system virtual prototyping (MSVP) is expected to reduce product design cycle with lower life-time cost, and has great significance to improve product quality. In virtue of this technology, engineers can simulate the characteristic of kinematics and dynamics of mechanical system on realistic condition by its digit model. In the dissertation, several key issues of developing an applicable MSVP including prototype modeling and dynamic numerical solution are addressed.At first, the basis concept and main task of MSVP are discussed, an architecture of software platform is proposed which can be compatible with various numerical solvers and support data exchange with different software. Furthermore, the associated relationships of model elements and the operation spreading are also concluded according to the model representation, an event-driven update mechanism is presented to implement model management and element relationship maintenance.Virtual prototype modeling is a basic problem in dynamic simulation of mechanism system. To facilitate model modification, integrate engineering design knowledge and support optimization analysis of model, a parametric modeling technique for MSVP is studied. According to the parameter trait, engineering constraint and associated relationship, a directed bi-partite graph is utilized to express parameter constraints. In terms of the partially ordered relationship of parameters, a solving sequence of reasoning is formulated. With the update mechanism, a parametric modeling method is proposed.To improve modeling efficiency and reuse of models, subsystem modeling techniques for complicated mechanism systems are addressed. According to the characteristic of model structure, virtual parts are presented to define the topological relationship among the subsystems, main control parameters are introduced to encapsulate the design knowledge and the internal parameters, the reference frames of subsystem are used to position. Consequently, a top-down hierarchical architecture is constructed. In terms of the hierarchical architecture and main control parameters, the algorithms are also presented to assemble the subsystems and solve the parameter constraints. Based on these techniques, the complicated model can be decomposed into smaller and simpler subsystems to construct and reuse.The kinematics and dynamics analysis is the essential function of MSVP. To improve the computationally efficiency of assembling equation of motion, a hybrid algorithm is presented by combination of global method and topological method. Firstly, the recursive formulations and Jacobian matrix are obtained by the kinematics relationship between reference frames of adjacent bodies, and then, according to the computational effect of cut joint, a weight graph is constructed to represent the topology of model. Based on this weight graph, the algorithms of the spanning tree generation with best efficiency and selection of base body are also proposed. Furthermore, the transformation formulation of Cartesian coordinates space to joint coordinates space is developed by model topology and recursive formulation, and an hybrid algorithm for assembling equation of motion is obtained.The main problems of dynamic numerical solution of mechanism system are discussed. Firstly, a perturbation on tangent-plane algorithm is proposed to distinguish the redundant constraint, and then a modified penalty numerical method is applied to dynamic analysis of mechanism system with singular configuration and topology change. To eliminate the constraint violation, the project correction of general coordinates is utilized. For the system with stiffness characteristic, an implicit integration scheme is adopt to directly discrete the equation of motion. To improve the computational efficiency, the residual of equation and approximated Jacobian matrix are also computed by the recursive formulations. In the end, the subsystem synthesis algorithm and divide-and-conquer algorithm is also studied to improve the computational efficiency and parallel of dynamic analysis.On the basis of the above proposed algorithms, a MSVP named InteDyna has been developed. Finally, the feasibility and validity for this research is testified using a whole vehicle modeling and simulation.