Approximate Evaluation of Sampled Model Parameter Matrices for Multivariable Continuous-time System

This paper shows that, under the condition of relatively small sampling period, the parameter matrices in sampled model of a multivariable linear dynamic system are mainly related to the first one or two coefficient matrices in the denominator and numerator polynomials of the original continuous-time model. Thus, some simplified formulas approximating the sampled model parameter matrices are then derived. The result obtained is useful for properly setting the initial values of recursive parameter estimation in self-tuning-type control so as 16 get a satisfactory closed-loop system response even during start-up of the control.