In many multichannel active noise and vibration control systems the controller is adapted to minimize the 2-norm of the error signals. This may, however, lead to a large spatial variance in the residual error. A method of achieving a more uniformly controlled error field using a weighted squared error strategy has previously been proposed, although the presented method of defining the error weighting parameters results in a very slow convergence rate. This convergence rate limitation has been overcome by the minimax algorithm which minimizes, in a least-squares sense, the maximum error signal at each iteration. However, due to the inherent switching in this algorithm, for fast convergence speeds it suffers from significant misadjustment and in a tonal control problem this introduces additional unwanted spectral components. In this paper an alternative method of minimizing the maximum error signal is proposed which uses an adaptive error-weighting matrix that is bounded and so avoids the slow convergence speeds previously reported. It is also shown that the proposed algorithm does not suffer from the same misadjustement problems shown by the minimax algorithm. The details of the proposed method are first outlined and then its performance is compared to the previously proposed methods through a series of time-domain simulations employing measurements of a physical system.
展开▼