Emilios K. Dimitriadis
John J. McCoy
School of Eng., Catholic Univ. of America, Washington, DC 20064
Most current transient signal control techniques are based on system identification and subsequent control of a discrete model. Estimation uncertainties, and also modal coupling via excitation, control, or damping determine the limits of successful control. The alternative suggested here, is based on a priori obtaining the ``effective'' dispersion characteristics of a chosen wave-propagation path. The dispersion induced by a complex waveguide may be determined by measuring a broadband pulse at the intended control-sensor and control-actuator locations. The technique of windowed Fourier transform is employed to deduce the ``effective'' dispersion curve(s) for said path. In a control situation, the sensed transient is computationally propagated in time-frequency space to the control location. The controller subsequently alters the time-frequency impedance characteristics of the control location to optimally achieve the desired control objectives. Numerical experiments are here described for two different beam structures. Laboratory experiments are suggested and their implementation discussed. Finally, issues of structural discontinuities, wave-mode conversion, and, very importantly, computational speed requirements are addressed.