An active noise control system based on reference signal decomposition

The performance of the traditional active noise control system will degrade severely when the target noise has both broadband and narrowband components. At present, many scientists have discussed this topic, but the problems that certain known conditions are required or high computational complexity still exist. In This article, we propose a reference signal decomposition normalized filtered-x least mean square (RSD-NFxLMS) algorithm. By means of linear prediction (LP), the narrowband signal will be separated from the reference signal, then the broadband and narrowband signal are processed by two parallel systems respectively. Besides, we proposed a coefficients-weighting (CW) method to balance the contradiction between convergence speed and steady-state error, which significantly improves the performance of the algorithm. The simulation results show that the algorithm can achieve the same effect as Hybrid ANC (HANC) system without using known conditions, especially in low signal-to-noise ratio (SNR) and low frequency conditions.