A model reference approach for implementing active structural acoustic control

Abstract

A model reference active structural acoustic control (ASAC) approach is investigated as a method for replacing microphone error sensors located in the acoustic field with accelerometers located on the structure. The suggested controller is the multi-input/multi-output adaptive LMS algorithm. As opposed to driving the response of the sensor located on the structure to zero with the control inputs, the response is driven to some pre-determined "reference" value corresponding to an a priori training cost function, which is originally chosen to minimize either the far-field pressure at a number of acoustic field points, or the supersonic region of the wave-number transform. In effect, the uncontrolled structure is adaptively modified to behave like the reference structure. Results indicate that the identical optimal control solution can be obtained when implementing the structural sensors, and the required number of sensors must simply equal the number of control actuators. In addition, for the example studied here, significant reduction in far-field sound radiation can be achieved even when errors of +/- 5% are present in the reference model. A considerable level of attenuation in sound can also be achieved for a band of frequencies within +/- 5% of the reference frequency used to obtain the models for creating the disturbance. Thus it appears that the model reference approach is reasonably robust.