Performance with regulation constraints (Q5926278): Difference between revisions

The authors consider a problem of achieving desired performance with a regulation constraint for continuous-time system of the form \[ \dot x = Ax + E_ww + E_dd + Bu,\quad e = C_ex + D_{ew}w + D_{eu}u, \] \[ z = C_zx + D_zu,\quad y = C_yx + D_{yw}w + D_{yd}d. \] The system describes the plant with state \(x\in\mathbb R^n\), control input \(u\in\mathbb R^m\), measured output \(y\in\mathbb R^p\), tracking error \(e\in\mathbb R^q\), and the controlled variable (for performance requirement) \(z\in\mathbb R^l\). The exogenous disturbance input \(w\in\mathbb R^s\) is generated by an exosystem with state-space realization \(\dot w=Sw.\) The variable \(d\) denotes an external disturbance. Typically, performance can be measured by the \(H_2\) or \(H_\infty\) norm or any other norm of a certain matrix. This optimization problem is solved indirectly by transforming it to an unconstrained optimization problem, however, for an auxiliary system instead of the given system. The auxiliary system is explicitly constructed from the data of the given system. A fundamental and significant issue that arises is whether the added regulation constraint compromises the achievable performance. In this regard, the authors show that when the performance is measured by the \(H_2\) norm of a transfer function matrix, there is no loss at all in the achievable performance. On the other hand, when the performance is measured by the \(H_\infty\) norm, there is however a certain loss or decay in the achievable performance. A very explicit expression for such a decay in performance is presented. This decay is expressed in terms of a static optimization problem.