Study On Adaptive Feedback Algorithm For The Active Noise Control Headrest

The feedback active noise control headrest has a simple structure because no reference sensor is required,and the short acoustical path makes it have relatively good noise reduction performance in middle and low frequencies.Non-adaptive feedback systems require the noise characteristics in advance and optimize the control filters offline,which cannot be applied to the situation where the noise spectrum changes.The adaptive feedback system is more flexible because it can track changes in noise and transmission paths in real time.However,the noise amplification caused by the waterbed effect,and the relatively large leakage factor used to avoid excessive lowfrequency amplitude of the control filter caused by the insufficient low-frequency response of the secondary sources due to the volume limitation of the active noise control headrest,will affect the noise reduction performance of the algorithm.In this thesis,adaptive feedback algorithms of the active noise control headrest are studied,and the new adaptive feedback algorithms for suppressing the noise amplification caused by the waterbed effect and the excessively large low-frequency amplitude of the control filter are proposed.The main work of the paper includes three parts:Firstly,in order to suppress the noise amplification caused by the waterbed effect of the feedback active noise control system and the excessively large low-frequency output of the secondary source,a frequency band constrained filtered-x least mean square(FcFxLMS)algorithm is proposed,which can constrain the amplitude of the control filter with different weights in different frequency bands by adjusting the magnitude frequency response of the spectrum shaping filter.The effectiveness of the algorithm is verified by simulations based on the measured transfer functions of an active headrest system.Compared with the existing algorithm,the performance of the proposed algorithm is the same but the computational complexity is significantly reduced.Secondly,in order to suppress the noise amplification caused by the waterbed effect when the noise spectrum changes and the location of the waterbed is unknown,a sensitivity constrained filtered-e least mean square(ScFeLMS)algorithm is proposed,which can track the changes of the noise spectrum and constrain the amplitude of the sensitivity function of the feedback system in real time,thereby constraining the noise amplification caused by the waterbed effect.Simulation results verify the effectiveness of the proposed algorithm in suppressing the noise amplification caused by the waterbed effect before and after the noise changes.Finally,for the case of where the in-vehicle active headrest is used for road noise control,the performance of the feedback system and the hybrid system based on the ScFeLMS algorithm and filtered-e least mean square algorithm are studied.Simulation results based on the road noise and the transfer functions of an active headrest measured in a new energy vehicle show that the noise reduction performance of the feedback system is approximately the same as that of the multichannel feedforward system,but the system complexity and cost are significantly reduced.Compared with the feedforward system and the feedback system,the hybrid system can further improve the noise reduction performance of active headrests.