The Investigation Of Active Noise Control Algorithms Regarding Different Secondary Path Characteristics

The noise level inside ship cabins is one of the important indexes to be considered in the process of ship design and the requirement from various institutions for the indicator has been increasing recently.The thesis adopts the active control technique to attenuate the lowfrequency noise inside ship cabins.Different control algorithms are investigated to improve the noise-reduction effect for control systems with nonminimum-phase,time-varying or nonlinear secondary paths.The details are as follows:Firstly,the thesis introduces the traditional active control algorithms and their characteristics are analyzed.Then the control performances of the algorithms are compared and discussed by simulations.The main factors affecting the noise-reduction performance in controlling of broadband noise are analyzed by attenuating the interior noise of a cabin as an example.The multi-channel feedforward adaptive algorithm is used in the experiment,and the acceleration signals of the structure are adopted as reference signals to reduce the noise inside the cabin.By comparing and analyzing the reference signals’ coherence,sampling frequencies and secondary paths,the main factors affecting the control performance of broadband noise are obtained and the simulation results are verified by experiments.Besides,challenges in reducing harmonic noise of engine systems are also discussed.Meanwhile,the influence of nonminimum-phase secondary paths on a broadband noise control system is analyzed in detail.The noncausal inversion of a practical nonminimum-phase secondary path is formulated and its influence on the noise-reduction performance is analyzed.Based on multiple coherence between reference signals and the undesired noise,a novel formulation for identifying primary paths with correlated excitation signals is presented and a causal optimal controller is proposed.The proposed controller can be used as an accurate predictor to estimate the maximumly achievable noise reduction and provide a reference to improve the control systems.The robustness of proposed algorithm is examined by varying the uncertainty of the identified primary paths.Finally,the performance of the proposed causal optimal controller is validated using the data measured in experiments.For time-varying secondary paths,the traditional adaptive algorithms can’t reduce the noise effectively and the modelling errors of secondary paths may even cause instability in control systems.To tackle with the problem,the thesis introduces a new online secondary path modeling method with variable step-size and self-tuning power scheduling for the injected random noise.The gain of the injected random noise and step size of the modelling filter are adjusted according to the modeling errors and convergence states of the three adaptive filters.Compared with previous methods,the proposed method improves the convergence speed and estimation accuracy for the active control system and the secondary path modeling,respectively.The performance of the proposed method is verified by simulations with different kinds of excitation signals.In addition,the nonlinearity of secondary paths in control systems is also one of the dominant factors deteriorating the noise-reduction performance.In order to improve noisereduction performance for nonlinear control systems,the thesis designs a feedback controller based on Lyapunov function.Firstly,a robust controller is designed with linear matrix inequality for the linear part of system containing uncertain parameters;then a backstepping method is used to force the nonlinear loudspeaker system track the obtained robust controller and estimate the unknown system parameters.Besides,the extended Kalman state observer is used to calculate system states in real time,and the states’ statistical errors are taken into account to correct the designed controller.The noise reduction of using the proposed controller in the interested frequency band is illustrated by simulations.Then a simple and feasible strategy to identify both linear and nonlinear parameters of loudspeakers is proposed and a nonlinear loudspeaker is identified in the laboratory.The subspace identification method is also used to identify the state space matrices corresponding to the noise transfer paths.The limitations of the proposed algorithm are illustrated based on the above identification results.In order to improve the generality of the nonlinear noise control algorithms,nonlinear digital filters are used to design controllers in the discrete domain.And a filtered-x least mean square algorithm using reweighted bilinear filters is proposed in the thesis and a generalized filter to model different nonlinear secondary paths is explored.Within the controller of an adaptive feedforward control system,the proposed algorithm uses trigonometric expansions to achieve high-order nonlinearities and combines the cross terms between the controller output,sine and cosine functions of the input with the reference signal.As a result,a satisfactory control result can be achieved with a shorter memory length of the controller.It is also found that the introduction of a reweighting factor in the weight-updating process can effectively reduce the interference produced by the small coefficients of the control filter and accelerate the convergence rate of the control system.The computational complexity of the proposed algorithm is analyzed and its control performance in terms of convergence rate and residual error is verified through simulations with different nonlinear secondary paths and excitation noises.Finally,a TMS320F28069 DSP controller is used to implement the proposed algorithm and the control performance is verified by experiments.