Research On Active Control Algorithm Of Vehicle Interior Noise Based On Improved Hybrid Structure

Living standards improving,cars have evolved from a means of transportation to a place for relax.To meet the demand for ride comfort,manufacturers have been paying more attention to the NVH(Noise,Vibration,and Harshness)performance.In-vehicle noise can be divided into narrowband engine order noise and broadband road excitation.Under lower driving speed,in-vehicle noise is mainly composed of engine order noise,which the traditional broadnarrowband hybrid ANC(Active Noise Control)system can eliminate;when the vehicle speed increases,the uncorrelated interference at the error microphone increases,and the feedforward microphone cannot completely obtain the reference signal of the target noise reduction point,so the feedforward-back hybrid structure ANC system has a better noise reduction effect.However,the traditional broadband-narrowband hybrid structure requires the rpm signal as input,which limits its noise reduction effect to the engine order noise.And its noise reduction ability is insufficient for uncorrelated sinusoidal noise and other narrowband components with higher energy.The traditional feedforward-back hybrid structure is prone to "waterbed effect",and the noise reduction ability of its feedback subsystem needs further improvement.In this paper,we will make some improvements to the traditional hybrid structure ANC system in order to improve the applicability and noise reduction performance of the system.In order to improve the defects of the traditional hybrid structure ANC system,a broadnarrowband mixed ANC system based on linear prediction(LP)approach for reference signal decomposition is proposed,which simplifies the SNC(Sinusoidal Noise Canceller)subsystem and synchronous signal generation subsystem.In order to verify the feasibility of the LP,a comparison is made with the SNC subsystem under two different types of broadbandnarrowband hybrid input(broadband-sinusoidal hybrid,broadband-narrowband Gaussian hybrid)operating conditions,respectively.In addition,a normalized filter coefficient weighting(CW)approach is proposed,in which each order of weighting coefficients can independently adjust the weighting factor according to its magnitude change when following the convergence of the system to achieve a faster convergence while maintaining a better steady-state performance.Meanwhile,in order to improve the noise reduction performance of the conventional feedforward-back hybrid ANC system,LP and CW methods are introduced into its feedback subsystem.Since the feedback subsystem mainly targets the noise and additional noise that the feedforward subsystem fails to completely eliminate,and its reference noise also presents certain mixed characteristics of broad and narrow bands,the LP approach can further improve the performance of the feedback subsystem under mixed noise input.In order to simplify the step-size selection process of ANC system,a parameter matching method based on Simultaneous Perturbation Stochastic Approximation(SPSA)algorithm is developed in this paper.On the basis of ensuring the steady-state performance and convergence speed of the system,the SPSA algorithm can quickly select the near-ideal optimal solution and improve the efficiency of parameter matching.In order to verify the superiority of the improved hybrid structure ANC algorithm,real vehicle data are collected and simulated for comparison.The results show that the improved broad-narrow band hybrid ANC system sacrifices a small amount of noise reduction at the engine order frequency point because it does not use the engine speed as the system input,but performs better at other frequency points and in a wider frequency band,especially for some narrow-band noise that is not related to engine noise or has higher energy.The improved feedforward-back hybrid ANC system not only enhances the noise reduction effect,but also has a good suppression effect on the "water bed effect" in the frequency band around 100 Hz.In order to demonstrate the applicability of the improved hybrid structure ANC system,the noise reduction effect of different ANC systems was compared under test bench conditions using real vehicle data.The results show that the improved broad-narrow band hybrid ANC system can improve the noise reduction by 0.5-1.5d B under different working conditions,and the improved feedforward-back hybrid ANC system also has a noise reduction improvement close to 0.5d B.Overall,the innovation of this paper is to propose a hybrid structure active noise control system that does not require rpm as input.Although this will somewhat weaken the noise reduction ability of engine order noise,it has greater improvement for the whole ANC system.This idea can further broaden the optimization direction of hybrid structure ANC system and has certain reference value.