Block recursive parallelotopic bounding in set membership identification (Q1129648): Difference between revisions

For the linear regression model \[ y(t)=\phi ^{'}(t)\theta ^{*} + e(t), \quad t=1,2, \dots \;, \] with the unknown-but-bounded noise \(e(t)\) (\( | e(t)| \leq \epsilon \), \(\forall t\)) the problem of estimation of the parameter vector \(\theta ^{*} \in R^n\) is considered. The techniques of parallelotopical approximation of polytopes are used to solve this set membership identification problem. The emphasis is laid on block processing with \(q > 1\) measurements (instead of one measurement) at each step of the recursive procedure and one obtains the optimal bounding parallelotope algorithm OBPq. Examples which illustrate the advantages of the OBPq method are provided.