Abstract:
The audio codecs in conventional active noise control systems can only use a high-speed single sampling rate, so the modeling filters and adaptive filters in the system require a longer filter length and the time reserved for iterations is very short, which leads to high computational complexity and difficulty in real-time online computation. To address these problems, a multi-sampling rate system is firstly proposed, which uses extraction and interpolation methods to provide lower sampling rates for active noise control system, reducing the computational complexity while increasing the time available for iterative computation. Based on the minimum mean square algorithm, the secondary sound channel is modeled and a multi-sampling rate double-channel feedback active noise control system is modeled too. Finally, an experimental platform based on seat headrest and digital signal processor is built to experimentally verify the effectiveness of the double-channel active noise feedback control system with multi-sampling rates. The experimental results show that for low frequency noise, the noise reduction system can achieve a noise reduction of about 15 dB in the corresponding frequency band, and the noise reduction is obvious.