An active noise control system, which minimizes acoustic energy density, can often become unstable due to significant changes in the acoustical environment in which it operates. Typically, the control system of interest, which is based on a modified version of the filtered-x LMS algorithm, is operated with a fixed set of system identification filter coefficients. These coefficients are determined off-line, prior to execution of the main control algorithm. The performance of the control system depends on the quality of the system model obtained through the system identification algorithm. Changes in the system transfer function and impulse response, due to various factors, including temperature, boundary conditions, and sensor/actuator location, have been investigated. Results will be presented, which give indications regarding what types of changes to the system have the greatest effect on the transfer function and impulse response, and how sensitive the control system is to those effects. The effects of changing sampling rates and system identification filter lengths on the quality of the system model have also been investigated and will be discussed.
Acoustical Society of America Journal
- Pub Date:
- October 2003