Digital simulation by partitioning and uncoupling the system model

Abstract

State variable simulation of linear time-invariant dynamic systems is reviewed with its advantages in accuracy and computational efficiency described. Then a technique for simulation of large order, nonlinear, time-varying dynamic systems is developed by partitioning the system state variable model and uncoupling the resulting subsystem models so that the linear time-invariant ones can be simulated by state variable solution. A simulation error introduced by this technique of partitioning and uncoupling is derived and analyzed in the form of a state variable model. Results are presented of a simulation error analysis utilizing the derived error state variable model for the simulation of a Saturn V attitude control system. In addition to computational savings, the technique of partitioning and uncoupling is shown by practical example to result in a modular simulation with flexibility to implement easily changes or additions. The digital simulations by this technique of two large-order dynamic systems, the LST (Large Space Telescope) fine pointing control system and a Shuttle Orbiter/Spacelab SEM (suspended experiment mount) precision attitude control system, are described; and simulation responses are included.