Active Noise Control Method Of Rotating Machinery Based On Signal Decomposition

The rotating machinery usually generates strongly low-frequency noise with respect to the rotational frequency of the shaft and its harmonics.The low-frequency noise not only accelerates the aging of equipment,but also brings serious noise pollution.Currently,active noise control(ANC)is an effective technique for a reduction of low-frequency noise.However,the noise will be non-stationary when rotating machinery operates under variable speed,especially for startup and shutdown.Moreover,the primary path and secondary path possibly have characteristics of nonlinearity and time-variety in practice.Accordingly,this paper has in-depth study on non-stationary noise control method of the rotating machinery under speed variation using the signal decomposition.The specific research contents and results are as follows.(1)To solve the problem of strongly non-stationary primary noise,an ANC control method based on real-time correlation empirical mode decomposition(EMD)is proposed.The primary noise is first decomposed into intrinsic mode functions(IMFs)by using the EMD.Then an indicator of periodic characteristic(IPC)based on the degree of cyclostationary is proposed to evaluate nonstationarity of primary noise and determine the number of decomposed IMFs.Subsequently,the correlation analysis between IMFs and primary noise is conducted for choosing useful IMFs that are controlled independently.The simulation results show that proposed method can effectively separate interesting components,with the IMF becoming more stable and easy to control.(2)Aiming at the problem of time-variety secondary path,a novel ANC method based on simultaneous perturbation and negative kurtosis is proposed.The feedforward subband ANC control structure is firstly established.By adding disturbance into subband controllers,the measurement value of cost function and gradient estimation can be obtained and used to update weight vector.Besides,the negative kurtosis of error signal is used as cost function to improve the convergence speed.Finally,we have a discussion on the convergence and computational complexity of proposed algorithm.Simulations have been carried out to show that proposed algorithm outperforms the online FxLMS,online MFxLMS and SPSA in noise reduction and convergence rate.(3)A nonlinear ANC method based on Chebyshev filter is proposed to overcome the problem of nonlinear primary path.A block-oriented approach is first used to establish nonlinear primary path model,and the primary noise distortion caused by the nonlinear primary path is then deduced and analyzed.Subsequently,a second-order Chebyshev filter is introduced to extend the decomposed IMFs,and the convergence and computational complexity of proposed algorithm are discussed.The simulation results show that proposed algorithm can effectively reduce primary noise and high-order harmonics caused by the nonlinear disturbance compared with the FxLMS,FSLMS and VFXLMS.