Study On Adaptive Algorithm In Active Noise Control And Application In Local Space Noise Reduction

Nowadays,noise becomes one of the most concerned problem in worldwide environmental pollution and makes more and more influence on human life.Among all kinds of noise,medium and low frequency noise is the one which causes large influence on human beings physically and psychologically.The technology of noise reduction is a significant way to solve noise problem and includes the passive noise control which adopts specific material structures and the active noise control(ANC)which adopts signal processing.However,the passive noise control is helpless to low frequency noise.Then ANC has become an efficient way to solve medium and low frequency noise.It is a hotspot research of noise control at present.ANC is made of the primary path and the secondary path which both generally adopt the least mean square(LMS)algorithm.The values of tap-length and step-size affect the performance of LMS algorithm in secondary path and primary path directly.Under the permissible tap-length range of hardware resource,too small value of taplength will cause a severe error of the secondary path identification and too large value of tap-length will also increase the computation and cause the misadjustment easily.Step-size has an impact on convergence and steady-state performance of primary path algorithm and secondary path algorithm.Therefore,research of taplength and step-size has a significant impact on the improvement of noise reduction effect in ANC.Currently,the adjustment algorithm research of step-size or tap-length is not deep enough.Some algorithms lack systematic and theoretical analysis and equip poor practicability.Moreover,application of ANC mainly concentrates on the duct and relevant research on the local three-dimensional scene is little.In this dissertation,the step-size and tap-length adjustment of LMS algorithm in ANC is studied deeply.And the research results are applied to noise reduction in local threedimensional space.Firstly,the structure and related algorithms of ANC are studied and analyzed in detail.The dissertation exactly connects some performance of adaptive LMS algorithm in primary path and secondary path with step-size and tap-length.The corresponding expressions of convergence performance,steady-state performance,robustness and computation have been studied and analyzed systematically.The typical variable tap-length algorithms and variable step-size algorithms are analyzed in detail.This part provides the theoretical guidance for researching on the tap-length or step-size adjustment of adaptive LMS algorithm in ANC.Secondly,the dissertation studies the tap-length adjustment of adaptive LMS algorithm deeply.This section provides three improved algorithms which mainly solve the step-size of variable tap-length,error width and large fluctuation problems in the fractional variable tap-length algorithm.For the step-size of variable tap-length,an improved algorithm which equips the adjustment strategy of step-size of variable tap-length is proposed.This algorithm can avoid the influence of system error and the problem of under-modeling.Then,to focus on the error width,a fractional variable tap-length algorithm with the strategy of variable error width is proposed.This algorithm not only adjusts error width according to convergence state reasonably,but also selects proportion factor dynamically.Thus,the limited function can restrict large fluctuation of tap-length without affecting convergence performance.These two improved algorithms are suitable to be applied in secondary path identification with low noise and small disturbance.Then it studies and analyzes the combination of step-size of variable tap-length and error width reasonably and uniformly.A fractional variable tap-length algorithm with mixed adaptive parameters is proposed.The algorithm also adds the method of dynamically choosing the proportion of the limited function.It optimizes the overestimation of tap-length and improves the performance of variable tap-length algorithm in high noise environment.Thirdly,the dissertation studies the step-size adjustment of adaptive LMS algorithm deeply.Aiming at the theoretical depth and practicability of variable stepsize algorithm,an iterative variable step-size algorithm is proposed.It is based on theoretical analysis and the reasonable expression is obtained by using the b asic formula deduction.The transient and steady-state performance is analyzed theoretically and experimentally.The results show that the proposed algorithm has good performance.It is suitable to be applied in primary path of ANC.Then,the dissertation studies and analyzes the combination of variable step-size algorithm and variable tap-length algorithm reasonably and uniformly.A hybrid algorithm which adjusts tap-length and step-size simultaneous is proposed.The theoretical analysis and experimental results show that the algorithm can obtain reasonable tap-length and step-size at the same time.It improves the steady-state performance further.Finally,the dissertation studies the ANC experiment in local three-dimensional space.The experiment platform of local three-dimensional scene in large enclosed space is built reasonably.All proposed algorithms are compared by the characteristic requirement of local three-dimensional space and then the appropriate algorithms are applied in the primary path and secondary path of ANC.Active noise elimination experiments are carried out by using various kinds of noise.The noise reduction effect is compared by the corresponding charts.The experimental results are consistent with the theoretical analysis.In secondary path identification,the proposed algorithm obtains the optimal tap-length under the permissible tap-length range of hardware resource and the reasonable error range of secondary path identification.It not only estimates the secondary path accurately,but also reduces the computational complexity and probability of misadjustment on the premise of guaranteeing the tracking convergence performance.In the primary path algorithm,the proposed algorithm improves the effect of noise reduction significantly.The research content of this dissertation improves the performance of algorithm in ANC and lays an experience for practical application.It has a certain significance to the development of ANC algorithm and local three-dimensional space noise control.