Point-type Broadband Active Noise Control Intelligent Algorithm For Low And Medium Frequency

With the development of society and the wide application of various electronic devices in daily life,noise is everywhere,and noise pollution,especially indoor noise pollution,has become one of the factors that seriously affect people’s daily life.Therefore,noise control has become a hot issue of widespread concern among current researchers.Active noise control(ANC)is an effective method for medium and low frequency noise.However,due to fixed step size,traditional Filter-x Least Mean Square(Fx LMS)algorithm has the problems of conflicting convergence speed and steady-state performance and weak anti-interference performance.Most of the existing active noise control has not been studied for indoor noise.Therefore,this dissertation takes indoor noise as an object for active noise control,based on the structure of feedforward broadband active noise control system,and proposes a variable step size algorithm based on genetic algorithm for tuning parameters and a variable step size algorithm based on interval type-2fuzzy set,which improve noise reduction performance of the system and enhance its anti-interference performance.The main research contents are as follows.Firstly,indoor noise is mainly composed of superimposed noise generated by electronic devices such as air conditioners,refrigerators and computers,which have the characteristics of wide band and low and medium frequency,and it is very difficult for current equipment to achieve noise reduction in the whole indoor space.Therefore,noise reduction range is limited to a point range,and the structure of point feedforward broadband active noise control system is studied.This dissertation proposes a variable step size Fx LMS algorithm based on improved sigmoid function to establish the relationship between error signal and step size,to solve the problem of conflicting convergence speed and steady-state performance of the Fx LMS algorithm and improve noise reduction performance of the system.Secondly,a three-parameter-based improved sigmoid function is constructed.Genetic algorithm with nonlinear search capability is used to realize the optimization of the three parameters in order to obtain best noise reduction level.At the same time,a maximum step size limit is introduced considering that sudden disturbance signals appearing in the real situation can lead to instability and even divergence of active noise control system.Simulation results show that the proposed variable step size Fx LMS algorithm has good noise reduction performance and anti-interference performance.Finally,this dissertation explores the nonlinear relationship between frequency characteristics of the noise and step size by frequency domain.Considering the uncertainty of step size in variable step size Fx LMS algorithm model,interval type-2 fuzzy set is applied to establish a relationship between frequency of the noise and step size.Therefore,this dissertation constructs a variable step size Fx LMS algorithm based on interval type-2 fuzzy set.According to the simulation experiments under different conditions,it is verified that the variable step size algorithm based on interval type-2 fuzzy set algorithm has better performance and anti-interference for medium frequency noise.There are 42 figures,12 tables and 76 references.