Robust adaptive algorithms for active noise and vibration control

Adaptive algorithms are described which simultaneously estimate and track the two transfer functions online. No random noise generator is required. Since both transfer functions are estimated, the resulting adaptive algorithms are much more robust than previous ones in that moving the transducer and/or the actuator around does not affect the performance much. Real-time implementation results for one of the algorithms presented are compared with results obtained using the LMS algorithm. It is found that the performance of the LMS algorithm depends greatly on the number of taps used, whereas the new algorithm is almost unaffected by this factor. This demonstrates the robustness of the new algorithm