Abstract:
The thesis presents a method of calculating the radiated sound power of vibrating structures based on the time domain estimation of acoustic radiation modes (ARMs). Each ARM is frequency-dependent, radiates power independent of the other ARMs and can be estimated in the time domain from measurements made at discrete sensor locations on the surface of the radiating structure. The individual ARM components are estimated digitally in the time domain using finite impulse response filters, which are designed to provide a best weighted fit to the ARMs in the frequency domain. The ARM amplitudes are estimated by filtering the vectors of measured velocities at points on the radiating surface with these ARM filters, before summing the product of the square of these amplitudes multiplied by the relevant ARM eigenvalues to estimate the radiated sound power. The method is described with reference to one and two dimensional radiators, namely a simply supported beam and a clamped plate. The results show that the sound power calculated from the proposed approach and from a frequency domain approach are comparable. Next, real time simulations of active structural acoustic control are performed using real-time ARM amplitudes as the cost function of the controller. The primary path is the path between the disturbance signal and the error signal, in this case, the ARM amplitude. The secondary path is from the controller output to the error signal. Two control strategies are considered here. The first one is nonadaptive feedforward active structural acoustic control, which is applied to a baffled beam. The controller transfer function is defined by the ratio of the primary path and secondary path frequency responses. This technique requires the control path to be accurately estimated using an FIR filter to get good attenuation. The second strategy is adaptive feedforward active structural acoustic control with reference to the baffled plate as the radiator. The controller is based on the filtered-x version of the adaptive LMS algorithm. Here two FIR filters are required to estimate the secondary and the control paths. To get good attenuation, the optimal locations of the control actuators are obtained using a swarm intelligent algorithm called Ant Colony Optimization. Finally, physical experiments are conducted to validate the findings.