| Multichannels adaptive active noise control (AANC) systems are applied commenly in the field of noise control to expand the spatial extend and enhance the reduction of sound. However, the corresponding computational load of multichannel control algorithms will become huge with the increase of channel numbers. This thesis focuses on the implementations and performances of multichannel fast algorithms based on filtered-X LMS algorithm in multichannel AANC systems.At first, the single- and multichannel Filtered-X LMS (FXLMS) algorithm are deduced and concluded respectively. The criterion of computational load of AANC algorithm is summaried with the features of the DSP controller in AANC systems, and the specific computational loads of FXLMS algorithm are counted. The effects of system parameters on the computational load are also studied.The basic idea and detailed classifies of fast algorithm in AANC are discussed. Two kinds of fast algorithms, i.e. structure improvement and coefficients fast update fast algorithms are concerned. The basic principles, implementations and operation procedures of all fast algorithms are specificly exponded. Computational loads of all fast algorithms are concluded and compared.The coefficients patial update fast algorithms are studied systematically. The static state patial update fast algorithms are summaried and developed, and a new kind of fast algorithm which called selective patial update fast algorithms is derived. In the case of multichannel system, a new multichannel fast algorithm is obtained by combining the strategy of selctive patial update and fast exact.Finally, the computational load, convergency performances and steady state characteristics of all fast algorithms are compared by computer simulations while modeling primary and secondary paths as linear minimum-phase low-pass filters. It is shown that the proposed fast algorithms have better overall performances than the others in multichannel AANC systems.
|
PDF Full Text Request