System Identification for Coprime Factor Plant Description with Bounded Error

One way to describe models with unstructured model errors in robust control is a coprime factor description with bounded coprime factors errors. In this paper a methodology is derived to identify these coprime factors with their bounds. The true process is assumed to be linear time invariant and the noise is assumed to be bounded in the frequency domain. With the use of R-parametrization an upperbound for the coprime factor errors can be derived and minimized in a convex setting. A simulation example shows that the algorithm easily converges and that the upperbound is correct.