Directional active noise control with a local minimax error criterion

Abstract

The traditional mean squared error (MSE) criterion can be formulated as a quadratic function of a vector of control filter coefficients, and it is easy to obtain optimal control filter coefficients. Although the MSE criterion can lead to noise reductions in the control area, an unpredictable directional residual sound field is generated. In this paper, we propose a method for multi-channel active harmonic noise control with a local minimax error criterion based on the Nelder-Mead algorithm, which leads to reductions at all error positions and greater reductions at controllable positions. Directional noise reduction experiments of two areas are presented for two different error criteria at discrete locations in an anechoic chamber. Compared with a system employing the traditional MSE criterion, the results show that an active noise control system with the proposed criterion can achieve extra reductions at specified locations and overall noise reductions at the same time. The present research offers some important insights into directional control.