A New Modified Wavelet based Adaptive Step Size for Improving Performance of Online Secondary Path Modeling in Active Noise Control Systems

In this paper a modified version of the Akthar’s method for online secondary path modeling in active noise control systems is proposed. The proposed method defines two variable step sizes for control filter and modeling filter. The control filter step size μm is adapted by using the variable threshold value Tj and the modeling filter step size μm is adapted by using a new modified variable step size (MVSS) in least mean square (LMS) algorithm. The step size of the modeling filter is varied in accordance with the ratio of power of the disturbance signals. In many ANC applications, online secondary path modeling method using white noise as training signal is required to ensure convergence of the system. The proposed method stops injection of white noise, when desired signal d(n) is equal to secondary signal u`(n) using wavelet denoising. The method uses two types of wavelet thresholds. One is modified soft threshold and other is soft threshoding. It is shown in the proposed ANC system with modified soft threshoding, in addition to soft threshoding, the residual error signal and the secondary path modeling error can be reduced at a fast convergence rate than Akthar’s method. The computer simulation results indicate the effectiveness of the proposed approach.