MFXRLS‐based adaptive feed‐forward controller implemented with velocity sensor identified by frequency response to improve actuator speaker performance in ANC systems

Abstract

To improve performance of a “dual voice coil” (DVC) actuator speaker in an active noise control (ANC) system, a speaker model that includes coupling dynamics and source of noise pressure is first derived to design an adaptive feed‐forward controller based on modified, filtered‐X, recursive‐least‐squares (MFXRLS) algorithm in this investigation. A novel velocity sensor measuring velocity of the speaker face is further developed by use of a frequency‐response method. Two transfer functions required for the velocity sensor are identified in two steps: (i) the adaptive feed‐forward controller is applied to keep speaker face velocity zero to identify the first transfer function; and (ii) the second transfer function is then obtained experimentally using the first transfer function. Performance of the established velocity sensor is similar to that of a Polytec OFV2100 laser velocity transducer. This velocity sensor is then incorporated with the adaptive feed‐forward controller to control the DVC actuator speaker in the ANC system. For a sinusoidal command input of frequency below 390 Hz, the controlled speaker acquires a unit‐gain magnitude and zero phase degree, showing that the controller can effectively reduce effects of the speaker dynamics, including coupling dynamics.