1982 Volume 18 Issue 4 Pages 349-356
In recent years, an adaptive observer which estimates the parameters and the states of an unknown plant has been actively examined. But most of studies are contributed only to a single-input single-output system. The extension of this procedure to the multivariable case seems to be more difficult because the canonical form for such systems has much more complicated structure.
This paper deals with the problem of desining an adaptive observer for a multi-input multioutput linear time invariant discrete system where only the inputs and outputs can be measured. In order to estimate unknown parameters of a plant, a scalar output representation is defined and the resultant multi-inputs single-output system is considered. For the estimation of output equation, two different algorithms are obtained by using a recursive least square estimation method.
The first approach makes use of estimated states for parameter estimations. In the second approach, the output equation is transformed into the linear combination of unknown parameters and known signales, and the signal is used for parameter estimation, and the iteration number of convergence is also determined. At the end of this paper, the results of digital simulation for a third-order system using two approaches are presented.