Simultaneous Online Modeling of the Secondary Path and Neutralization of the Feedback Path in an Active Noise Control System
Regular paper
Hochschule Offenburg
Wednesday 3 june, 2015, 08:40 - 09:00
0.7 Lisbon (47)
Abstract:
This paper proposes a new method for modeling the secondary path and
neutralizing the feedback path for a feedforward single channel active noise
control (ANC) system in a duct, during its online operation. Using white noise
with fixed variance for this purpose is the most widely used method but in the
final stage the noise adds up to the residual error and reduces the
performance of the ANC system. In the proposed method a gain scheduling
strategy is used to control the variance of the injected white noise. The
variance is in turn controlled by the convergence of feedback path
neutralization filter and secondary path modeling filter. Instead of the
traditional Least Mean Squared (LMS) algorithm, the Normalized Least Mean
Squared (NLMS) algorithm is used in the proposed method. This ensures
robustness of the system in practical situations where the power of the signal
can vary any time. The algorithm is tested in a MATLAB/Simulink environment.
Custom Simulink blocks like Fx_NLMS block were designed using Level-2 S-
Functions from Simulink. Simulations show that the proposed method ensures
reduced steady state error at the error microphone and provides an excellent
steady state performance even if the power of input noise is varied.