Active engine order noise equalization control combined with virtual microphones

Songming Qi,Yu Wang,Qiang Ge,Sifa Zheng
2018-01-01
Abstract:Low-frequency noise is a significant factor that influences the acoustical comfort inside vehicles. The adaptive feedforward control system is usually utilized to mitigate low-frequency noise, which may introduce the problem of spatial interference of error microphones of control system with the control area. The solution of this problem is using the virtual microphone technique. In view of the strong nonlinear characteristic of the sound field inside the vehicle, a nonlinear virtual microphone estimation method is proposed in this paper. The back propagation neural network is utilized to estimate the transfer function of acoustic signals in different positions inside the vehicle. Firstly, the characteristic of vehicle sound field is analyzed theoretically. Then the estimation model of nonlinear virtual microphone combined with neural network is established. The inputs of this model are two physical microphone signals and the engine speed signal, while the outputs are two virtual microphone signals. Integrating this model with the engine order noise equalization control model, the engine order can be adjusted independently by selecting different combinations of gain factors corresponding to the center frequencies of the equalized control reference signals. Through experimental verification, good control performance can be achieved using nonlinear virtual microphone for different control modes.
What problem does this paper attempt to address?