Corba-based Distributed Multi-body System Sensitivity Analysis

Distributed simulation, analysis and optimization are important fields of study in the co-design of multibody dynamical systems. Sensitivity analysis is important in the study of dynamic optimization of multibody system dynamics for it provides useful information for the optimal design and re-analysis.Based on the results published in recent years and adopting the technologies such as distributed object, computer network, multibody system simulation and database, a software environment for the sensitivity analysis of multibody system dynamics is set up in this paper. The main work includes:Adopt VisBroker. ORB product by Inprise Company, as intermediate to construct distributed computation emironment based on CORBA software. Adopt ACCESS DATA BASE to administer the related data. Choose C++Builder and Java as pre-developing tools. The background serving programs are written using Fortran language. Image simulation is carried out by applying OpenGL. The core of setting up the distributed computation environment is the data transfer bus. The requests are sent to the server though the bus. and the server sent the results of computation back to the end of client though the same bus.Based on the general mathematical models of multibody system dynamics such as the kinematics models of multibody dynamical systems, the ordinary differential equation models and the differential algebraic equation models of dynamics, general forms of the sensitivity equations of the systems are deduced by using direct differentiation method and adjoint variable method. For the mathematics model of differential/algebraic equation system, the penalty function is introduced in numerical computation, and Runge-Kutta method and Newton-Raphson method are used respectively in solving theordinary differential equations and nonlinear equation systems. The process and the results are carried out under CORBA environment.Based on above distributed simulation software of multibody system dynamics, three examples of distributed sensibility analysis, double pendulum system, planar four-bar mechanical system and slider-crank system, are numerically dealt with, the results of which indicates that the methods raised in this paper are correct. User of the softwaresend the basic parameters required in sensitivity analysis at the client's terminal. Other works such as the automatic creation of mathematics models, numerical computation and data management, and return of the simulation results are carried out automatically at the terminal of the server.The above work sets up a foundation for the distributed optimization analysis and optimization design of the computer added complicated mechanical systems.