Abstract
In this paper, we propose a method for identification of continuous-time transfer function models directly using sampled input-output data. In the proposed method, input-output derivatives are obtained by outputs of filters, and the transfer function can be represented as an auto-regressive model by using the derivatives. This method differs from the traditional parameter identification methods based on the least-squares, because measurement noise is considered in the criterion explicitly and the identification problem can be formulated as an optimization problem with constraint. The optimal solution is presented in this paper. The effectiveness of the method is verified through an example of a linear motor identification.
