Research On Partial Iteration And Variable Tap-Length Algorithm Of Multi-Channel Active Noise Control System

Active Noise Control(ANC)realizes the purpose of noise control through the active cancellation of the sound wave emitted by the signal source.With the rapid development of social industry and economy,people have higher and higher requirements for noise reduction,and multi-channel ANC system was born.With the increase of the number of channels,the computational complexity of the system increases sharply,and the conventional Fx LMS algorithm has been difficult to meet the requirements.It is a common practice to reduce the computational complexity of the system by partial iterative algorithm.The convergence performance of the algorithm affects the noise reduction performance of the whole ANC process,so it is extremely necessary to analyze the convergence performance of the partial iterative algorithm.It is worth mentioning that partial iterative algorithm only iterates a subset of the whole signal,which will lead to slower convergence of the entire ANC process,which is also a problem worth solving.Firstly,this paper discusses the noise reduction performance of some iterative algorithms on the same platform.Common partial iterative algorithms reduce the computational complexity of the system by modifying the cost function or partial iterative error signals and partial iterative filtering reference signals,such as Scanning Error algorithm,Minimax algorithm and Mixed Error algorithm.This paper introduces a new partial iteration algorithm,Bipolar Minimax,which can be partially iterated error signals or partially iterated filter reference signals,reducing the computational complexity of the system from two dimensions.This paper conducts comparative simulation experiments on(1,4,4)and(2,6,6)systems respectively.The simulation results show that Bipolar Minimax algorithm has significant advantages over other partial iterative algorithms in noise reduction performance.Secondly,the convergence performance of partial iterative algorithm is analyzed.The convergence boundary and steady-state characteristics of some iterative algorithms are deduced in detail,and the correctness of the theoretical derivation is verified by simulation experiments.The simulation results show that under correct convergence step size,partial iteration algorithms always converge in ANC process,and Bipolar Minimax algorithm can converge almost the same control filter weight vector as Fx LMS algorithm in general convergence direction.Finally,in view of the negative effects of partial iterative algorithms,a new variable tap-length algorithm is proposed.In the early stage of ANC process,the control filter taplength is iterated to obtain faster convergence speed,and in the process of convergence,the control filter tap-length is transformed to obtain high convergence performance.Simulation results show that the method is effective.