NOVEL DESIGN OF FEEDFORWARD ANC SYSTEM WITH SNR PROCESS GAIN
WU Ying-zi · Zhendong yu chongji · 2007
In some special applications of active noise cancellation (ANC) system, some useful signals may exist in the noise field, for example, the cases of vehicles (sound signals and engine noises) and ships (target signals and self-noises of ships), where the traditional ANC system doesn't work well. In view of this situation, a novel system structure is proposed in which the reference input of ANC system is pretreated using the technology of independent component analysis (ICA). Benefiting from the new system structure, the location of reference sensors can be more flexible and the signal-to-noise ratio (SNR) and correlation between new reference input and signals are lower compared with those in the traditional ANC system. Consequently, better performance of system is obtained, such as faster convergence speed, higher SNR process gain and lower level of signal distortion. Another important advantage is that the capability of selective noise cancellation is achieved. The theoretical analysis is done and computer simulative experiments are also accomplished using real data. The experiment results show that the proposed scheme is more effective than the traditional ANC system for these especial applications.