A method stably working feedback type active noise control system for preventive panel of sound leakage

In this paper, we propose a method capable of providing the maximum noise reduction effect without adjusting the step size. The feedback type active noise control system estimates the coefficients of the noise control filter by applying the filtered-x algorithm to the steepest descent method, for example, the least mean square (LMS) algorithm. The steepest decent method, however, stipulates the upper limit of the step size guaranteeing the stable estimation according to the auto-correlation function of the reference noise. Our target is the reduction of non-stationary noise, such as speech signal and traffic noise, whose auto-correlation function is continuously changing. The filtered-x algorithm moreover processes the reference noise with the feedback control filter, which changes the auto-correlation function according to the impulse response of the feedback path. We then propose to apply the sub-recursive least square (sub-RLS) algorithm not requiring the inverse matrix operation to the estimation of the coefficients of the noise control filter. The Sub-RLS algorithm can estimate the coefficients independently of the auto-correlation function, and provides almost the same convergence speed as that of RLS algorithm. We finally verify using computer simulation that the sub-RLS algorithm successfully reduces the reference noise with a fixed step size.