Active impedance control of linear one-dimensional wave equations (Q4955134)

The authors study impedance control for the one-dimensional classical wave equation \[ \left({A\partial^2\over\partial x^2}+{B\partial^2\over\partial x\partial t}+{C\partial^2\over\partial t^2}\right) P(x,t)= Q(x,t),\;x\in (0,1),\;t> 0,\;B^2> 4AC, \] where the impedance is the ratio of velocity and force, voltage to current, etc. Taking the Laplace transform, and replacing time \(t\) by the frequency variable \(s\), the authors introduce homogeneous boundary conditions in the frequency domain for \(\widetilde P(0,s)\), and \(\widetilde P(1,s)\), where \(\widetilde P\) denotes Laplace transform of \(P\). The Green's function for these equations, which are rather complex have been copied from \textit{A. G. Butkovskii's} handbook [Green's Functions and Transfer Functions Handbook, Chichester (1982) (translation from Russian)]. The control problem posed here is to find a control satisfying some impedance matching conditions. For example one may wish to control an acoustic noise wave. The control objective is a complete blockage, that is a total reflection. Or else, one may wish to have no reflection, so that a wave passes right through some boundary. In this impedance matching technique the complete knowledge of boundary conditions is not needed to derive the control law. The feedback controllers display infinitely may poles located along the imaginary axis, as one would expect in the case of loss-free wave propagation, and the system displays a lack of internal stability. A control law is derived for active noise control in ducts, to illustrate this theory.