Research On Active Road Noise Control Method And Its Performance Optimization In Pure Electric Vehicles

Pure electric vehicles are the most rapidly developing and mature of all new energy vehicles and are currently the mainstay of new energy vehicles.Pure electric vehicles use electric motors instead of fuel engines and lose the masking effect of engine noise,resulting in a more prominent road noise problem compared to traditional fuel vehicles.At low and medium speeds,road noise has become the main noise of pure electric vehicles.Therefore,the effective control of road noise in pure electric vehicles is of great significance to improve their NVH performance and enhance driving comfort.The energy of road noise is mainly concentrated in the low-frequency range below 500 Hz,which is not well controlled by traditional passive noise control techniques.On the contrary,active noise control techniques have been widely proven to be more capable of controlling low-frequency noise.Therefore,road noise control methods based on active noise control have good prospects.Nowadays,many companies and institutes at home and abroad are investing in the development of active road noise control systems,which have become a new trend in the field of intelligent NVH.The design and development of active road noise control systems involves acoustics,digital signal processing,control science,and many other fields.It is a complex system engineering.To further improve the noise reduction capability of the active road noise control system,this paper carries out a study on the active road noise control method and its performance optimization,starting from the two key technical problems of reference signal selection and control algorithm optimization to lay the foundation for the practical engineering development of active road noise control systems.The main contents of the paper are briefly described as follows:Firstly,a study of the reference signal selection method for the active road noise control system is carried out.Data acquisition of vibration signals and interior noise signals of a joint venture brand pure electric SUV is conducted.Multiple coherence analysis is carried out on the collected data to determine the high coherence frequency range of the vibration and noise signals.The eigenvalue curves of different vibration signals are obtained by means of principal component analysis,and the number of reference signals used is determined accordingly.In consideration of the actual noise reduction band of the active road noise control system,the Fisher information matrix is constructed using the multiple coherence functions of the vibration and noise signals,and then the reference vibration signal set used in the active road noise control system is determined according to the constructed Fisher information matrix so that the selected reference signal set can have a higher multiple coherence with the noise signals in the vehicle.Secondly,a study on the performance optimization of the adaptive filtering algorithm for the active road noise control system is carried out.The method of generalized filtered reference signal matrix and filter coefficient matrix decomposition is proposed.It is pointed out that the coefficients of the same adaptive filter can be updated separately by a number of the same adaptive algorithms using different parameters or a number of different adaptive algorithms,in order to combine the advantages of different parameter conditions or different algorithms.In addition,a step-size self-search strategy is designed.By observing the changing conditions of the designed parameters in real time,the step size is adaptively adjusted in order to eliminate the disadvantages of traditional algorithms that require pre-set step size parameters.Based on the characteristics of the road noise transfer path,a step-size weighting matrix is designed to assign different weights to the adaptive filters corresponding to different vibration signals,so that the performance of the adaptive algorithm can be fully utilized.A purely time-delayed secondary path transfer function is developed to replace the traditional estimated secondary path transfer function in the updating process of the filter coefficients,in order to eliminate the possible adverse effects of the estimated secondary path transfer function’s magnitude response on the adaptive algorithm.By combining the above strategies,a self-searching step size and matrix decomposition for normalized filtered-x least mean square(SS-MD-NFxLMS)algorithm is obtained.A simulation model of the algorithm is developed and simulations are carried out under different operating conditions.The results demonstrate that the designed algorithm significantly improves the noise reduction capability of the active road noise control system.Thirdly,a study on the computational complexity optimization of the adaptive filtering algorithm for the active road noise control system is conducted.By introducing a delayless subband structure and designing a partial subband update strategy,the adaptive algorithm only processes the noise signals within the target band,thus significantly reducing the computational cost required by the adaptive filtering algorithm.Combining the previously designed step-size self-search strategy and the purely time-delayed secondary path transfer function,a delayless partial-updated subband,self-searching step size,and matrix decomposition for normalized filtered-x least mean square(DPUS-SS-MD-NFxLMS)algorithm is proposed.The computational cost of the DPUS-SS-MD-NFxLMS algorithm is analyzed and the prototype low-pass filter is designed.A simulation model of the algorithm is developed for simulation analysis.The simulation results show that the designed algorithm can still achieve significantly better performance than the conventional algorithms at low computational complexity.Finally,based on the spatial active noise control test platform built by the research group,an experimental study of the active road noise control algorithm is carried out using pre-collected vibration signals and interior noise signals to verify the noise reduction performance of the proposed adaptive algorithms.The actual noise reduction performance of the active road noise control algorithms is analyzed and the proposed adaptive algorithms are verified.The experimental results show that the proposed algorithms can achieve significant noise reduction in the frequency band of approximately 50-450 Hz,with noise reduction in the key frequency band of 1.77～11.61 dB.The proposed algorithms also show significant performance improvement over the conventional algorithms,proving the superiority and advancement of the proposed algorithms.