A study on active noise control based on preinverse adaptive system with bias free

In an active noise control (ANC) system, Filtered-x algorithm is widely used as feedforward adaptive system. However the algorithm has a possibility of making the ANC system unstable, because of the modeling error of secondary path. In order to solve the problem, a pre-inverse type ANC is proposed in this paper. In this system, an adaptive filter estimates the inverse transfer function of secondary path and the weights of the adaptive filter are copied to the filter, which is inserted before the secondary path. Then a primary path is estimated by another adaptive filter. Therefore, the primary and secondary paths are controlled independently and the proposed ANC improves the noise reduction ability. Unfortunately, the pre-inverse type ANC degrades the estimation accuracy of the secondary path due to bias. Thus, the bias free adaptive algorithm with square sum of the correlation is also proposed. The adaptive algorithm takes advantage of the independence between the input signal and disturbance.