Research On Active Noise Control System In Vehicle Based On Multi-gradient Adaptive Algorithm

Since the idea of active noise control（ANC）was put forward,ANC technology has been widely concerned and applied due to its good suppression effect on lowfrequency noise,and has gradually approached our lives,especially in passenger cars.However,the noise environment in the car often changes in real time according to the driving environment and contains complex interference noise.In addition,the convergence speed and computational complexity of the algorithm have always been the focus of improving the algorithm.The computational complexity of an algorithm determines its practicability and rationality.Excessive computational load may cause the controller to overheat and even affect the efficiency of the entire system.Therefore,seeking an algorithm with strong adaptability,fast convergence speed and low computational complexity has always been an important direction for the improvement of ANC algorithm.On the other hand,the Filter-x least mean square（FxLMS）algorithm is the basic algorithm of ANC,and the accuracy of secondary path identification severely limits its use.Although offline secondary path identification can reduce the complexity of the algorithm,the secondary acoustic path is often time-varying,and offline secondary path identification cannot provide accurate secondary path transfer coefficients in real time.The online secondary path identification method has better real-time performance and can meet the requirements of FxLMS algorithm for accurate secondary path transfer function.Therefore,the development of online secondary path modeling algorithms is also of great significance.Based on the above problems,this paper studies the active noise control algorithm based on the multi-gradient method and the secondary channel online identification method.The specific research contents are as follows:In this paper,a multi-gradient FxLMS（MGD-FxLMS）algorithm is proposed to optimize the update of the weight vector of the control filter.The algorithm determines the base point of the conversion gradient from the instantaneous absolute value of the error signal and the ratio of the power recursive estimate of the error signal to the power recursive estimate of the control filter output signal,and it uses the gradient of the FxLMS algorithm,the gradient of the FxLMF algorithm and the gradient of the FxatanLMS algorithm as the three gradient directions of the proposed algorithm to update the weight vector,respectively,in order to improve the convergence speed,improve the noise reduction effect and reduce the computational load.By comparing the performance of the proposed algorithm with existing algorithms through computational complexity analysis and computer simulation analysis,the computational complexity of the MGD-FxLMS algorithm is determined according to the characteristics of the noise,and the MGD-FxLMS algorithm has significant noise reduction advantages for power-varying analogue reference noise,strong impulse analogue signals and actual in-vehicle noise.In this paper,an online secondary path modelling algorithm with auxiliary white noise is proposed for secondary path identification.The online secondary path identification method is optimized in terms of both the auxiliary noise gain scheduling strategy and the variable step size strategy of the modelling filter.The proposed algorithm has certain advantages in terms of computational complexity.The parameters of the proposed algorithm are selected by particle swarm optimization algorithm,and the proposed algorithm is demonstrated to outperform other algorithms in terms of noise reduction effect and modelling error by computer simulation.According to the requirements of arctangent auxiliary noise gain secondary path online identification method,the main hardware equipment is selected,the corresponding hardware-in-the-loop test platform is designed and built.This platform can switch the secondary path by switching the error microphone.The superiority of the online secondary path identification algorithm was verified using noise from a vehicle travelling at a constant speed.The identified secondary paths trended in line with the results of offline identification.The noise reduction effect is better than the offline identification method after the secondary path change.