Study On Simplified Active Noise Control Algorithms Without Secondary Path Modeling

Active noise control(ANC)is a method for attenuating unwanted disturbances by introduction of controllable secondary sources,whose outputs are arranged to interfere destructively with the original primary sound.For most ANC algorithms,it is usually necessary to identify the secondary path transfer functions between the control sources and the error sensors.ANC algorithms without secondary path modeling are a promising way to circumvent the influence of the secondary path,and have attracted interest of many researchers recently.The algorithms can be divided into three categories.(1)Method assuming the secondary path as a delay.This method is difficult to be realized in a practical system due to the unknown length of delay and infeasible priori information regarding filter parameters.(2)Updating several adaptive filters simultaneously.(3)Optimal updating direction searching method with subband adaptive filter architecture.The ANC algorithm without secondary path modeling is comprehensively investigated in the dissertation,with special attention paid on is implementation on the subband and frequency domain adaptive filter structures:Firstly,the simplified ANC algorithm without secondary path modeling is proposed.The delay reference signal is obtained by filtering the reference signal with the direct path delay filter.The orthogonal delay compensated reference signal is obtained by filtering the reference signal with the Hilbert filter.Two reference signals are generated in each subband,and 4 update directions are searched.The cost function is defined as the sum of the weighted squared subband errors,and the adaptive filter coefficients are updated in the time-domain.The proposed algorithm has the merits of lower computational complexity and simple implementation.Simulations and experiments are carried out to demonstrate the feasibility of the proposed algorithm in a duct with 3 varied acoustic end conditions.Secondly,the "subband + FFT" method is adopted in the traditional frequency domain subband ANC algorithm.A new frequency-domain ANC algorithm without secondary path is proposed,whose main benefits are that the frequency bins are obtained by FFT method,and that the reference signal and error signal are divided uniformly into a number of narrow bands by a special matrix to replace the polyphase method.The frequency-domain controller filter coefficients are updated independently with optimizing of its corresponding update direction.For narrowband active noise control,narrowband time domain ANC algorithm without secondary path modeling is proposed.The frequency bins can be obtained by the DFT method,and the 2N length complex exponential vector is divided into much smaller parts.The effective frequency bins are uniformly divided into several frequency bin groups,and the adaptive filter coefficients are updated in the time domain.The proposed algorithm has low computational complexity in variety length of adaptive filter.Finally,the DCT transform ANC algorithm without secondary path modeling is proposed.The reference signal and error signal frequency bins are extracted by DCT transform and divided uniformly into several frequency bin groups.There are two update directions for each frequency bin,0°and 180°.The adaptive controller filter coefficients are updated in each frequency bin with its corresponding update direction.the computational complexity can be reduced since the DCT-based algorithm is realized with all real calculations.