APPROXIMATE INTEGRATOR-GAINS FOR DISCRETIZING DIFFERENTIAL RICCATI EQUATIONS

摘要

Several new discretization methods are proposed for a differential Riccati equation and viewed from the point of a discrete-time integrator gain. The obtained discrete-time models, called the TL, TN, and TE models, are based on approximations of this gain, which plays a key role in the discretization and parameter identification of nonlinear systems. From numerous simulations, it was found that, in general, the TL and TN discrete-time models have good accuracy and remain stable even for large sampling intervals, whereas the TE model and the well-known forward-difference-model have poor accuracy and easily become unstable as the sampling interval increases. Overall, the TL model with a free parameter chosen to be zero usually gives the best performance among those tested in this study.