An efficient algorithm for output-error model reduction†
The paper presents an algorithm for reduced order modelling of linear discrete-time systems, using an output-error criterion. Let e(z) be the z-transform of the error between the impulse response of the given transfer function and that of the reduced-order approximate model: \(e(z)=b^ 0(z)/a^ 0(z)-b(z)/a(z)\) where both \(a^ 0(z)\) and a(z) are assumed to be stable. The cost function \(V(a,b)=e^ Te\) is expressed in terms of three matrices, each of them satisfying a matrix Lyapunov equation. These relationships are used to derive a formula for the gradient vector \(\partial V/\partial a\) and \(\partial V/\partial b\). This procedure is shown to be efficiently computable and requires a number of operations proportional to \(N^ 2\), where N denotes the order of the given system. A Fletcher-Powell method utilizes the gradient vector in order to compute the reduced - order model parameters by minimizing the error function V(a,b). The algorithm provides stability monitoring at no extra computational cost. Also exposed are some aspects about the model reduction of multi-input multi-output systems. The last section demonstrates the efficiency of the algorithm by three illustrative examples.
- Approximation of discrete linear systems
- scientific article; zbMATH DE number 3746882 (Why is no real title available?)
- scientific article; zbMATH DE number 3466802 (Why is no real title available?)
- scientific article; zbMATH DE number 3482173 (Why is no real title available?)
- scientific article; zbMATH DE number 3296351 (Why is no real title available?)
- Inverses of Toeplitz Operators, Innovations, and Orthogonal Polynomials
- On another approach to the Schur-Cohn criterion
- A new algorithm for \(L_ 2\) optimal model reduction
- Efficient input-output model representations
- A symbolic matrix decomposition algorithm for reduced order linear fractional transformation modelling
- A framework for reduced order modeling with mixed moment matching and peak error objectives
- A new approach to reduced-order modelling
- Model reduction using the output equation error
- Hybrid method for model reduction with time delay
- Second-order algorithm for optimal model order reduction
- Reduced gradient computation in prediction error identification
- ESTIMATION OF AUTOREGRESSIVE MOVING-AVERAGE MODELS VIA HIGH-ORDER AUTOREGRESSIVE APPROXIMATIONS
- A method for linear model reduction using the input-independent equation error
This page was built for publication: An efficient algorithm for output-error model reduction†
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3321923)