Algorithmic Issues in the Synthesis of Dissipative Systems
The authors of this paper study the problem of synthesis of dissipative systems. They assume the system dynamics to be specified in the most general form: a latent variable representation. The attention is focused on the synthesis of dissipative system behaviors. Necessary and sufficient conditions for the existence of such behaviors were given in some previous works. Concrete algorithms for the verification of these existence conditions are given in this paper. For the case that these conditions are satisfied, constructive algorithms to compute such controlled behaviors are obtained. A behavior \({\mathcal B}\) of a linear differential system is a subspace \({\mathcal B}\subseteq{\mathcal D}=C^\infty (\mathbb{R},\mathbb{R}^w)\) such that, for some polynomial matrix \(R\in\mathbb{R}^{g\times w}[\xi]\), we have \({\mathcal B}=\{w \in C^\infty(\mathbb{R}, \mathbb{R}^w)\mid R(\frac{d}{dt})w=0\}\). The set of such behaviors is denoted by \({\mathcal L}^w\). Usually, a subset of \({\mathcal L}^w\) is considered, namely the controllable behaviors. A controllable behavior is a behavior in which for any two of its elements there exists a third element which coincides with the first one on the past and with the second one on the future. \({\mathcal L}^w_{\text{cont}}\) denotes the subset of controllable behaviors. For \(w\in{\mathcal B}\in{\mathcal L}^w\) it is possible to choose some of the components of \(w\) to be any function in \(C^\infty (\mathbb{R},\mathbb{R})\). The maximal number of such components is called the input cardinality of \({\mathcal B}\) and is denoted by \(m({\mathcal B})\). Given a nonsingular matrix \(\Sigma\in\mathbb{R}^{w\times w}\) such that \(\Sigma= \Sigma^T\), a behavior \({\mathcal B}\in {\mathcal L}^w_{\text{cont}}\) is called \(\Sigma\)-dissipative if \(\int_\mathbb{R} w^T\;\Sigma\;w\,dt \geq 0\) for all \(w\in{\mathcal B}\cap{\mathcal D}\). \({\mathcal B}\in {\mathcal L}^w_{\text{cont}}\) is called \(\Sigma\)-dissipative on \(\mathbb{R}_-\), if \(\int^0_{-\infty} w^T\;\Sigma\;w\,dt\geq 0\) for all \(w\in{\mathcal B}\cap {\mathcal D}\). \({\mathcal B}\) is strictly \(\Sigma\)-dissipative if there exists an \(\varepsilon>0\) such that \(\int_\mathbb{R} w^T\;\Sigma\;w\,dt\geq\varepsilon \int_\mathbb{R} w^Tw\,dt\) for all \(w\in {\mathcal B}\cap{\mathcal D}\). Strict dissipativity on \(\mathbb{R}_-\) is defined analogously. The main purpose of this paper is to obtain algorithms concerning the following two problems. Let \(\Sigma=\Sigma^T \in \mathbb{R}^{w\times w}\) be nonsingular and let \({\mathcal N},{\mathcal P}\in{\mathcal L}^w_{\text{cont}}\) be such that \({\mathcal N}\subseteq{\mathcal P}\). The first problem is the problem of dissipative system synthesis: find \({\mathcal K}\in {\mathcal L}^w_{\text{cont}}\) such that: (i) \({\mathcal N}\subseteq{\mathcal K}\subseteq {\mathcal P}\), (ii) \({\mathcal K}\) is \(\Sigma\)-dissipative on \(\mathbb{R}_-\), (iii) \(m({\mathcal K})=\sigma_+(\Sigma)\), where \(\sigma_+(\Sigma)\) is the number of positive eigenvalues of \(\Sigma\). The second problem is the problem of strictly dissipative system synthesis: find a \({\mathcal K}\in{\mathcal L}^w_{\text{cont}}\) that is strictly \(\Sigma\)-dissipative on \(\mathbb{R}_-\), instead of just \(\Sigma\)-dissipative on \(\mathbb{R}_-\). A \({\mathcal K}\) having these desired properties is called a controlled behavior.
- A parametrization for dissipative behaviors
- The strict dissipativity synthesis problem and the rank of the coupling QDF
- Dissipative distributed systems
- DISSIPATIVE DYNAMICAL SYSTEMS IN A BEHAVIORAL CONTEXT
- Control of plant-wide systems using dynamic supply rates
- Fast modes in the set of minimal dissipation trajectories
- Dissipative systems synthesis: A linear algebraic approach
This page was built for publication: Algorithmic Issues in the Synthesis of Dissipative Systems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4459998)
