Adaptive Algorithms For Active Control Of Impulse (-like) Noises

Active noise control (ANC), ranking with the passive noise control, is a well-known approach to solve the problem of noise pollution in the environments. After the tireless efforts of researchers for decades, ANC system, especially the adaptive ANC system achieves many concrete and encouraging progresses and the theoretical framework and research method of ANC have been established. However, with the developments of research and the scope of engineering applications, some difficulties are encountered and there are many urgent issues in need of study and solution.The adaptive algorithms in the adaptive ANC system are investigated in this dissertation and the target is to seek the algorithms which have fast convergence speed, low computation complexity, good steady-state performance and robustness for the practical ANC system. The main contributions of this dissertation are summarized as follows:For the adaptive feed-forward control of a stable distribution noises, the cost function which is defined as the logarithmic transformation square of error signal is introduced and the FxlogLMS algorithm is derived. The FxlogLMS algorithm does not require any prior knowledge or estimation of characteristic exponent and exhibits good stability for all the cases with different characteristic exponents. Numerical simulations demonstrate the effectiveness of the proposed algorithm.The M-Estimator based algorithm family is examined in the context of active impulse-like noise control, where several existing algorithms are reformulated into the algorithm family and two new members, i.e. the Fair and Cauchy algorithms, are proposed. The general characteristics of the algorithm family in terms of robustness and computation load are analyzed, and the convergence condition of the proposed new algorithms is developed mathematically. Simulations are performed to verify the effectiveness of the algorithms and it is shown that the proposed algorithms compare favorably with the competitor algorithm in terms of performance, complexity, and threshold selection, and the Cauchy algorithm exhibits better performance under a stable distribution noise condition.An improved algorithm named PFNLMS is proposed which features that the filter coefficients are processed by a post filter and then utilized for adaptation. It is shown that the PFNLMS algorithm exhibits inherent flexibility since it can balance the convergence speed and steady state performance by tuning the parameter in the post filter. Because the cost function of the PFNLMS algorithm is defined as the summation of the squared Euclidean norm of difference between the currently updated filter coefficients vector and all past filter coefficients vector subject to the constraint imposed on the adaptive filter output, The PFNLMS algorithm is very suitable for the active control of impulsive noise since it directly limits the dynamic range of the adaptive filter coefficients and prevents the heavy fluctuation of the filter coefficients. Simulations compare the performance of the proposed algorithm with the existing algorithms. In addition, a procedure of selecting an appropriate algorithm from the family for practical applications is illustrated by simulations of active control of recorded pile driving noise.The widely used adaptive feedback active noise control system is based on the internal model control (IMC) structure where the reference signal is regenerated by synthesizing the error signal and the secondary signal filtered with the estimation of the secondary path. A simplified adaptive feedback active noise control system which uses the error signal directly as the reference signal is presented and the advantages of the simplified adaptive feedback active noise control system are the low computational load and ease of implementation. The convergence and stability is investigated and the common observations in practical active noise control system support that the simplification is reasonable. The simulations and experiments of active noise control system in a long duct for narrow band noise suppression show that the performance of the simplified adaptive feedback active noise control system is comparable to that of the adaptive IMC feedback system.