Research on Active Noise Control System Based on FBFULMS Algorithm

Convergence speed, steady-state performance and noise reduction performance of active noise control systems have been issues we need to address. In particular, bin-Normalized Frequency Domain Block Least Mean Square (NFBLMS) algorithm will have a biased steady-state solution when adaptive filter length is insufficient, while Modified Frequency Domain Block Least Mean Square (MFBLMS) algorithm and the mixed algorithm composed of NFBLMS and MFBLMS have slow convergence speed and poor noise reduction performance. In this paper, Frequency Domain Block Filtered-U Least Mean Square (FBFULMS) algorithm structure is proposed, which ensures both stability of system and fast convergence speed by means of secondary path modelling. Simulation results show that when input excitation is compressor noise data, the proposed FBFULMS algorithm achieves a noise attenuation of up to 26 dBA with 39 iteration steps. At last, XILINX Artix7-100T FPGA is used as the core control module for hardware design. Its measured results show that this method effectively attenuates the low and medium frequency noise in the target area of compressor, especially at the error microphone with a measured noise reduction of up to 20dBA and a convergence time of about 60µs.