Design And Implementation Of Software For Acoustic Feedback Suppression In Train Public Address System

The train public address system plays an important role in the train operation,but the acoustic feedback phenomenon caused by the microphone and loudspeaker located in the same space poses a serious threat to the normal operation of the train public address system.Early acoustic feedback suppression methods,such as frequency shifting,phase modulation,gain reduction,and array filtering,all have some defects,and the effect of acoustic feedback suppression is not ideal.With the development of adaptive theory,people have proposed an adaptive acoustic feedback cancellation method based on adaptive filtering for the acoustic feedback problem.The adaptive feedback cancellation method acts on the initial stage of acoustic feedback establishment,and can completely eliminate the influence of the acoustic feedback signal in theory,so it is sought after by people.In this thesis,aiming at the phenomenon of acoustic feedback in train public address system,the software for acoustic feedback suppression in train public address system based on adaptive feedback cancellation method was designed and implemented.First of all,in order to solve the correlation problem in the adaptive feedback cancellation method,three kinds of decorrelation methods were introduced,and the adaptive filtering principle involved in the adaptive feedback cancellation method and the existing adaptive filtering algorithm were briefly introduced.Secondly,in view of the problem that the existing Modified Least Mean Square(MLMS)algorithm cannot obtain low steady-state error while achieving rapid convergence,the Variable Step Size MLMS(VSSMLMS)algorithm was proposed by adopting the Variable Step Size scheme.Based on MATALB platform,VSSMLMS algorithm,Normalized Least Mean Square(NLMS)algorithm and MLMS algorithm were simulated and compared in two conditions of fixed acoustic feedback path and variable acoustic feedback path.The simulation results show that in both cases,the acoustic feedback suppression performance of VSSMLMS algorithm is superior to the other two adaptive filtering algorithms.Then,based on VSSMLMS algorithm,the software for acoustic feedback suppression in the train public address system was designed and implemented on the embedded system.Finally,the test platform of the train public address system was built,and the software designed in this thesis was tested.Test results show that the software for acoustic feedback suppression in the train public address system can effectively suppress the howling caused by the acoustic feedback.