Optimal Placements And Feedback Gains Of Sensors And Actuators For Active Noise Control

Active structural acoustic control (ASAC) is a main noise control strategy in the low to medium frequency range. It can effectively reduce the acoustic radiation from vibrating structures by applying secondary forces/sound sources on their surfaces. There mainly exist two widely applied methods in ASAC. One is vibration modal-based control, which places the actuators at the positions corresponding to the largest displacements of the vibration modes. The other is acoustic radiation modal-based control, which places the actuators at the positions corresponding to the largest normal velocities of the acoustic radiation modes. By orthogonalizing the normal velocity and the acoustic radiation impedance matrix, it can significantly reduce the acoustic radiation power. Whichever method is used, positions of actuators and magnitude of secondary forces/sound sources play a decisive role to the control effect. Therefore, positions of actuators and their feedback gain must be select reasonably to achieve an ideal result.Despite the fact that the above mentioned methods can sometimes achieve a fairly good control result, there still exist several problems. (1) The positions of the secondary forces/sound sources are determined a priori by experience, and the magnitudes are optimized independently instead of optimizing the positions and the magnitudes simultaneously. (2) Cost of control is not considered, sometimes resulting in an unrealistic secondary forces/sound sources. This thesis provides a thorough research on the above problems. Positions and feedback gains of the actuators are optimized simultaneously by the Davidon-Fletcher-Powell method. Control cost is introduced into the object function to make the result more reasonable. Finally, the feedback optimization strategy is employed to active control with random excitations. Numerical simulation demonstrates that the proposed approach works well not only for the single point harmonic excitation case, but also for the multiple point random excitation case.